In USA more than 12.5% of all infants are born preterm. Approximately 75% of all perinatal deaths occur among these preterm infants. Preterm infants are frequently very low in birth weight (VLBW) and receive multiple red blood cell (RBC) transfusions. These transfusions pose increased risk of infections and other complications. Since erythropoietin (EPO) stimulates RBC production, EPO treatment of VLBW infants has received attention as a modality for reducing transfusions in this group. The overall hypothesis of this work is that treatment optimization of EPO of anemia in preterm infants requires a comprehensive knowledge of the behavior of RBC and the pharmacokinetic/pharmacodynamics (PK/PD) relationship between EPO and erythropoiesis. Under that overall hypothesis, the specific aims were: 1) To describe erythropoiesis dynamics in preterm infants, 2) To determine and explain the variability in the response to EPO in preterm infants, 3) To evaluate newborn sheep as an experimental model for erythropoiesis in preterm infants, 4) To test the hypothesis that RBC lifespan is shortened under acute hypoxic stress conditions, 5) To test the hypothesis that EPO receptor (EPOR) pool size increases under hypoxic stress conditions and the change in EPOR pool size can be predicted using EPO clearance measurements, 6) To describe the effect of EPOR pool size changes on erythropoiesis kinetics. A model that describes erythropoiesis dynamics in preterm infants as a function of the plasma EPO concentration is presented in Chapter 2. In Chapter 3, several covariates are tested for their ability to identify infants with good EPO responsiveness. The lamb is also tested as an animal model for the erythropoiesis in preterm infants (Chapter 4). In Chapters 5-7, the effect of hypoxic stress conditions on RBC survival was explored defining the relation between the efficacy of EPO and survival of RBC produced as a result of EPO administration. RBC lifespan measurement methods are reviewed in Chapter 5. In Chapter 6, a new methodology for the measurement of RBC lifespan under stress conditions is developed. This new methodology is applied in Chapter 7 to explore the effect of hypoxic stress conditions on the survival of RBC. The study presented in Chapter 8 is undertaken to investigate changes in both EPOR pool size and EPO clearance under hypoxic conditions. An erythropoiesis model that accounts for change in the EPOR pool size under stress conditions is presented in Chapter 9. Analysis of erythropoiesis dynamics in preterm infants demonstrated that a three fold increase in the amount of RBC produced by preterm infants is possible by EPO administration. This emphasizes the potential of using EPO for the management of anemia in preterm infants. Covariate screening identified gestational age as a potential marker for the responsiveness to EPO treatment. PD analysis results in lambs demonstrated similarities between lambs and preterm infants in different erythropoietic characteristics such as sensitivity to EPO in producing RBC, Hb production rate before birth and blood volume. Survival analysis demonstrated that RBC lifespan is not shortened under acute hypoxic conditions Analysis of EPOR mRNA level demonstrated an up regulation of EPOR level under stress conditions accompanied by a parallel increase in EPO clearance. EPOR up regulation under stress conditions level was incorporated in a PD model presented in Chapter 9. The developed model provides a framework for optimizing EPO dosing. Accordingly, an optimal dosing strategy should in general maximize the interaction between EPO and EPOR. Specifically, EPO should be administered when the number of EPOR are close to maximally up-regulated.

<p>Artemisinin is the parent compound of a novel family of antimalarials. Repetitive administrations of artemisinin to both healthy volunteers and malaria patients have been shown to result in decreased plasma concentrations of the compound, most probably due to an autoinduction of different CYP450 enzymes. The aim of this thesis was to investigate the clinical pharmacokinetics and efficacy of different dosage regimens of the drug, and study the kinetics of the enzyme induction. Moreover, the putative interaction of the compound with blood components was investigated in vitro. </p><p>Artemisinin was found to distribute into red blood cells, competing with oxygen for binding to hemoglobin. The compound was stable in plasma and, in contrast to previous reports, did not bind to red blood cell membranes. </p><p>To circumvent the logistical and ethical problems associated with plasma sampling, suitability of saliva as substitute was investigated. Moreover, due to the large number of collected samples, an HPLC method, enabling a direct injection of saliva and plasma samples, was developed. </p><p>Saliva artemisinin concentrations were found to correlate with its unbound plasma levels, making saliva a suitable body fluid for pharmacokinetic studies of the compound. Based on saliva samples, artemisinin was shown to exhibit a dose-dependent kinetics and efficacy in malaria patients, with a possible sex-effect on the metabolism of the compound during the first treatment day. Moreover, the time-dependent kinetics of the compound was observed in both malaria patients and healthy subjects. A physiological approach was utilized to model the autoinduction in the latter group. A model with a feedback mechanism of enzymes was able to describe the data, with estimations of the half-lives of induction (3.15 hrs) and elimination of enzymes (32.9 hrs), as well as pharmacokinetic parameters of artemisinin. </p><p>In conclusion, artemisinin was found to exhibit a fast induction of enzymes, with time- and dose-dependent drug kinetics and dose-dependent antimalarial efficacy. </p>

<p>Cefuroxime is a renally eliminated antibiotic used against a variety of different bacterial infections. The pharmacokinetics (PK) for cefuroxime was studied in 97 hospitalized patients using population analysis. To be able to measure cefuroxime in human serum a new sensitive analytical method was developed using mass spectrometry detection. The method was validated and shown to be sensitive and selective. Cystatin C was found to be a better covariate for cefuroxime clearance compared to the traditionally used creatinine clearance (CLcr). This relation might be useful when designing dosing strategies for cefuroxime and other renally eliminated drugs. </p><p>The time-courses of the biomarkers C-reactive protein (CRP), serum amyloid A (SAA), interleukin-6 (IL-6) and body temperature were studied for the first 72 hours of cefuroxime treatment and was related to the duration of illness previous treatment with cefuroxime and to time to step-down of treatment. When duration of illness was short, CRP and SAA were showed increasing levels. None of the biomarkers could be used to differentiate between early or late step-down of therapy.</p><p>By use of known PK and pharmacodynamic (PD) principles, dosing strategies based on CLcr for cefuroxime were estimated using minimization of a risk function. The risk function was constructed with the aim to expose patients to cefuroxime concentration above minimum inhibitory concentration (MIC) for 50 % of the dosing interval and to minimize the amount of drug administered in excess to reach the aim. Based on evaluation using wild type MIC distributions for <i>Escherichia coli</i> and <i>Streptococcus pneumoniae</i> improved dosing strategies were selected.</p><p>In vitro experiments were performed exposing <i>Streptococcus pyogenes</i> to constant concentration of benzylpenicillin, cefuroxime, erythromycin, moxifloxacin or vancomycin. A semi-mechanistic PK/PD model characterizing the time-course of the antibacterial effect was developed using all data simultaneously. Internal validation showed the model being predictive and robust. </p>

<p>Pharmacometrics uses models based on pharmacology, physiology and disease for quantitative analysis of interactions between drugs and patients. The availability of software implementing modern statistical methods is important for efficient model building and evaluation throughout pharmacometric data analyses.</p><p>The aim of this thesis was to facilitate the practical use of available and new statistical methods in the area of pharmacometric data analysis. This involved the development of suitable software tools that allows for efficient use of these methods, characterisation of basic properties and demonstration of their usefulness when applied to real world data. The thesis describes the implementation of a set of statistical methods (the bootstrap, jackknife, case-deletion diagnostics, log-likelihood profiling and stepwise covariate model building), made available as tools through the software Perl-speaks-NONMEM (PsN). The appropriateness of the methods and the consistency of the software tools were evaluated using a large selection of clinical and nonclinical data. Criteria based on clinical relevance were found to be useful components in automated stepwise covariate model building. Their ability to restrict the number of included parameter-covariate relationships while maintaining the predictive performance of the model was demonstrated using the antiarrythmic drug dofetilide. Log-likelihood profiling was shown to be equivalent to the bootstrap for calculating confidence intervals for fixed-effects parameters if an appropriate estimation method is used. The condition number of the covariance matrix for the parameter estimates was shown to be a good indicator of how well resampling methods behave when applied to pharmacometric data analyses using NONMEM. The software developed in this thesis equips modellers with an enhanced set of tools for efficient pharmacometric data analysis. </p>

<p>The worldwide increase in antibiotic resistance is a concern for public health. When the appropriate antibiotic dosage is determined, the priorities are efficacy and toxicity. The aim of this thesis was to gain knowledge about the most efficient dosing regimens in order to minimize the emergence and selection of antibiotic-resistant mutants. We also wanted to assess the impact of antibiotic selective pressure and host to host transmission for the dissemination of resistance.</p><p><i>Escherichia coli </i>bacteria with different levels of cefotaxime susceptibility were competed in an in vitro kinetic model, demonstrating a complex selection of low-level resistance influenced e.g. by the time duration of selective concentrations and the rise of new mutants. We also constructed a mathematical model incorporating biologically relevant parameters and showed its usefulness when assessing the risks of resistance development.</p><p>When <i>E. coli </i>populations with pre-existing fluoroquinolone-resistant mutants were exposed to simulated serum concentrations, several currently used doses of fluoroquinolones clearly enhanced the development and selection of resistance. </p><p>The mutant prevention concentration (MPC) was measured for several <i>E. coli</i> isolates with different fluoroquinolone susceptibilities, and because of fluctuating antibiotic concentrations in the human body, the pharmacokinetics was considered when evaluating MPC. Results indicate that the area under the serum concentration time curve in relation to the MPC may be a useful predictor for emergence of resistance.</p><p>In the commensal flora of healthy human couples we noted a high frequency of trimethoprim-resistant <i>E. coli.</i> There was also an extensive sharing and transmission of <i>E. coli</i> clones. Treating the female with trimethoprim reduced the number of intestinal <i>E. coli</i> which might have facilitated the transmission from the male partner. These findings suggest that the rate of transmission is high and effectively contributes to the spread of both susceptible and antibiotic-resistant <i>E. coli</i> in intrafamilial settings.</p>

<p>Enzyme induction is a process whereby a molecule enhances the expression of enzymes. If the affected enzymes are involved in the elimination of a drug, this may result in a drug interaction. Induction is therefore of major concern during drug development and in clinical practice. </p><p>The induction process depends on the half-life of the induced enzyme, the pharmacokinetics of the inducing agent, and the relationship between the inducer’s concentration and the induction stimulus. The aim of the conducted research was to investigate these key aspects of enzyme induction and the consequences that induction has for substrate elimination.</p><p>Successful investigations of the induction process presuppose the existence of appropriate methods for the estimation of the metabolic activity. Enzyme activity measurements can be conducted in tissues with low enzyme content using the analytical method presented here. </p><p>A model was developed describing the changes in the pharmacokinetics of clomethiazole and its metabolite NLA-715, that are attributable to carbamazepine induction. The consequences of the induction was explained using a mechanistic approach, acknowledging food-induced changes in the blood flow to the liver, and interpreting in vitro generated metabolic information.</p><p>The time course of the induction process was examined in two investigations. In the first of these, the pharmacokinetics of the autoinducing drug phenobarbital and its effect on several enzymes were described in rats. This was accomplished by integrating the bidirectional interaction between drug and enzymes in a mechanistic manner. In the final investigation, the time course of the increase and cessation in enzyme activity was studied in healthy volunteers treated with carbamazepine. This investigation allowed the half-lives of CYP3A and CYP1A2 to be estimated. </p><p>The key aspects of the enzyme induction process have been examined using mechanistic induction models. These novel models improve the understanding of the induction process and its consequences for substrate elimination.</p>

The influence of hyperlipidemia on the pharmacodynamic and pharmacokinetic aspects of lipophilic drugs was explored. The antiarrhythmic, amiodarone, and the antifungal, ()-ketoconazole, were used as model drugs. Experimental hyperlipidemia was induced in rat using poloxamer 407 and two sensitive novel HPLC assays were developed. In a multiple dosing study, hyperlipidemia increased amiodarone plasma concentrations, heart concentrations and electrocardiographic changes. The amiodarone heart uptake could not be totally attributed to its unbound fraction, where the cardiac very low density lipoprotein receptors seemed to play a role in the uptake of bound drug. Amiodarone liver metabolism in presence and absence of hyperlipidemia was studied using isolated primary rat liver hepatocytes. The metabolism of amiodarone was lower in hepatocytes isolated from hyperlipidemic than those from normolipidemic rats. Hyperlipidemic serum resulted in a decrease in amiodarone metabolism and when coincubated, the expected decrease in unbound fraction seemed to resulted in greater inhibition of metabolism. ()-Ketoconazole showed stereoselectivity in its pharmacokinetics in rat with (+)-ketoconazole showing higher plasma concentrations than its antipode. This was attributed to its higher protein binding. There was no difference in the total bioavailability of the two enantiomers. Ketoconazole enantiomers exhibited nonlinear pharmacokinetics. In normolipidemic rat plasma ketoconazole enantiomers were more than 95% bound to lipoprotein deficient fraction. Hyperlipidemia resulted in shifting both enantiomers 20% to very low density and low density lipoprotein fractions. In a pharmacokinetic assessment, hyperlipidemia was found to increase ketoconazole enantiomer volume of distribution. Moreover, the stereoselectivity ratios of most pharmacokinetic parameters were changed. After oral dosing, the uptake of (-)-ketoconazole was significantly decreased. Since ketoconazole is used as a potent CYP3A inhibitor, alteration in liver concentrations of (-)-ketoconazole, the more potent inhibitory enantiomer, could decrease its CYP inhibitory potential. Hyperlipidemia potentiated the CYP-mediated interaction between ketoconazole and midazolam with significantly higher midazolam AUC and lower clearance. This was attributed to the inhibitory action of ketoconazole and the effect of hyperlipidemia on the binding of midazolam. Hyperlipidemia was found to unexpectedly decrease midazolam unbound fraction in plasma. In conclusion, the findings could explain some unexpected dose versus effect outcomes in hyperlipidemic patients receiving amiodarone or ketoconazole. / Pharmaceutical Sciences

In the past, hyperlipidemia (HL) has been shown to affect the pharmacokinetic and pharmacodynamic properties of lipophilic drugs extensively bound to lipoproteins, including halofantrine (HF). The present thesis examined the effect of HL on the biodistribution, metabolism and electrocardiographic (ECG) effects of HF in the poloxamer 407 rat model of HL. The HL state caused unexpected changes in the distribution of HF enantiomers. In contrast to plasma, concentrations of desbutyl-HF (DHF) were much higher in highly perfused tissues. Following in vitro incubation of racemic HF and DHF, HF and DHF enantiomers shifted from the lipoprotein deficient fraction to triglyceride-rich fractions in HL plasma. No significant differences were noted in HF metabolism between NL and HL liver microsomes. It appears that both reduced plasma unbound fraction and lipoprotein associated directed uptake of lipoprotein-bound drug by tissues play roles in enantiomer biodistribution. In everted gut metabolism, formation of DHF enantiomers was inversely proportional to bile concentration whereas addition of lipids in the presence of bile caused a significant decrease in DHF:HF ratio of (-)-enantiomers. Pre-treatment of rats with peanut oil had no significant effect on DHF formation in the incubated sacs or microsomal preparations, nor did it affect the expression of intestinal CYP450. The above results were consistent with previous in vivo evaluations showing that the DHF to HF ratio was decreased by the ingestion of a high fat meal. In the ECG study, HL by itself had no effect on the ECG. In HL rats, plasma but not heart concentrations of the HF enantiomers were significantly higher compared to NL rats. DHF did not impart significant ECG prolonging effects after HF administration. The unbound fraction of HF was the controlling factor for drug uptake by the heart. Despite the lack of difference in HF heart concentrations, the QT and QTc intervals were significantly higher in HL compared to NL rats, suggesting a greater vulnerability towards HF induced QT interval prolongation in the HL state. The HL serum resulted in decreased metabolism of HF enantiomers in the isolated primary rat hepatocytes. Studies with LLC PK1 and NRK 52E cells showed that HF is not a substrate of P-glycoprotein transporters. / Pharmaceutical Sciences

Artemisinin is the parent compound of a novel family of antimalarials. Repetitive administrations of artemisinin to both healthy volunteers and malaria patients have been shown to result in decreased plasma concentrations of the compound, most probably due to an autoinduction of different CYP450 enzymes. The aim of this thesis was to investigate the clinical pharmacokinetics and efficacy of different dosage regimens of the drug, and study the kinetics of the enzyme induction. Moreover, the putative interaction of the compound with blood components was investigated in vitro. Artemisinin was found to distribute into red blood cells, competing with oxygen for binding to hemoglobin. The compound was stable in plasma and, in contrast to previous reports, did not bind to red blood cell membranes. To circumvent the logistical and ethical problems associated with plasma sampling, suitability of saliva as substitute was investigated. Moreover, due to the large number of collected samples, an HPLC method, enabling a direct injection of saliva and plasma samples, was developed. Saliva artemisinin concentrations were found to correlate with its unbound plasma levels, making saliva a suitable body fluid for pharmacokinetic studies of the compound. Based on saliva samples, artemisinin was shown to exhibit a dose-dependent kinetics and efficacy in malaria patients, with a possible sex-effect on the metabolism of the compound during the first treatment day. Moreover, the time-dependent kinetics of the compound was observed in both malaria patients and healthy subjects. A physiological approach was utilized to model the autoinduction in the latter group. A model with a feedback mechanism of enzymes was able to describe the data, with estimations of the half-lives of induction (3.15 hrs) and elimination of enzymes (32.9 hrs), as well as pharmacokinetic parameters of artemisinin. In conclusion, artemisinin was found to exhibit a fast induction of enzymes, with time- and dose-dependent drug kinetics and dose-dependent antimalarial efficacy.

Pharmacometrics uses models based on pharmacology, physiology and disease for quantitative analysis of interactions between drugs and patients. The availability of software implementing modern statistical methods is important for efficient model building and evaluation throughout pharmacometric data analyses. The aim of this thesis was to facilitate the practical use of available and new statistical methods in the area of pharmacometric data analysis. This involved the development of suitable software tools that allows for efficient use of these methods, characterisation of basic properties and demonstration of their usefulness when applied to real world data. The thesis describes the implementation of a set of statistical methods (the bootstrap, jackknife, case-deletion diagnostics, log-likelihood profiling and stepwise covariate model building), made available as tools through the software Perl-speaks-NONMEM (PsN). The appropriateness of the methods and the consistency of the software tools were evaluated using a large selection of clinical and nonclinical data. Criteria based on clinical relevance were found to be useful components in automated stepwise covariate model building. Their ability to restrict the number of included parameter-covariate relationships while maintaining the predictive performance of the model was demonstrated using the antiarrythmic drug dofetilide. Log-likelihood profiling was shown to be equivalent to the bootstrap for calculating confidence intervals for fixed-effects parameters if an appropriate estimation method is used. The condition number of the covariance matrix for the parameter estimates was shown to be a good indicator of how well resampling methods behave when applied to pharmacometric data analyses using NONMEM. The software developed in this thesis equips modellers with an enhanced set of tools for efficient pharmacometric data analysis.