Abstract The global aims of this thesis were: to evaluate if an individualised dose strategy for enoxaparin, based on lean body weight and renal function, resulted in a reduction in the prevalence of bleeding and bruising events when compared to conventional dosing; to further understand the dose-exposure-response relationship for enoxaparin using population pharmacokinetic-pharmacodynamic (PKPD) modelling. This thesis comprises seven chapters: an introduction to the current knowledge and literature pertaining to low-molecular-weight heparins (LMWHs), in particular enoxaparin; five research chapters; and a discussion. Each of the five research chapters consists of a manuscript that has been published in, accepted or submitted for peer review in a scientific journal. Preceding each chapter is a synopsis of the important features of the publication. Supplementary information that supports the findings of the publication, but could not be included in the publication, is provided at the end of the chapter. Appendices relevant to each chapter are located at the end of the thesis. Chapter one is the introduction to this thesis. It commences with an overview of the LMWHs, their mechanism of action, customary uses, licensed doses and adverse effects. There is a brief introduction to renal function and body composition; physiological factors that influence the disposition of LMWHs. As much of this thesis is centred on defining the dose-exposure-response relationship for enoxaparin, there is a critique of the existing literature relevant to each segment of this relationship, namely: dose-exposure, exposure-response and dose-response. To conclude this chapter there is a review of pharmacostatistical models and population modelling, followed by an appraisal of population PK and PKPD models that have previously been developed for enoxaparin, including the two key publications that are critical to this thesis. These two papers were the first to fully describe the dose-exposure relationship for enoxaparin in subjects with renal impairment and obesity. It is from these studies that the individualised dosing strategy, explored throughout this thesis, was developed. The specific aims of the five research chapters are then stated. Chapter two describes a confirmatory, randomised controlled trial (RCT) to compare an individualised dose of enoxaparin to conventional, label based dosing. The RCT was conducted at a major tertiary teaching hospital over an 18 month period. The primary endpoint was the prevalence of overt bleeding events and the secondary endpoint a combination of bleeding or major bruising events. A time-to-bleeding event analysis (Kaplan-Meier) was used and markers of effectiveness such as mortality and readmission were assessed out to 30 days post recruitment. Bleeding and bruising data, along with anti-Xa (aXa)-concentrations were collected for use in additional research described in chapters four and five. Chapter three details the evolution, progression and contemporary knowledge of drug dosing based on body composition and focuses on dosing in obese subjects with cardiovascular disease. The concept of dose-individualisation is explored in this chapter with reference to the methods used to normalise drug exposure across the spectrum of body compositions. Subsequently, there is a review of body size descriptors, such as lean body weight, that are used to scale dosing in the obese. Enoxaparin is used as a motivating example, with reference to data presented in Chapter two of this thesis. There is also a discussion about the type of research designs that are required to maximise information about PK parameters. This chapter was published within a book chapter which was intended for clinical practitioners in the discipline of cardiology. Chapter four is focused on the development and evaluation of population PK and PKPD models to describe the time-course of effects for enoxaparin. A population PK model linked to a proportional-odds model was used to describe the severity of an adverse event as a function of exposure and demographic variables. The final model was used to explore the likely occurrence of bleeding and bruising events in patients with obesity and / or renal impairment dosed using either the individualised or conventional dose strategies from Chapter two. Chapter five describes a study that was undertaken to evaluate the ability of the individualised dosing strategy to achieve and maintain aXa-concentrations within the therapeutic range (500 to 1000 IU L-1), by comparison to conventional dosing. As the confirmatory study focused on the prevalence of adverse events there was no assessment of the therapeutic capability of the dose strategies however, as aXa-concentrations were collected using a sparse design during the confirmatory study, the two dose strategies could not be compared using observed data. Therefore, the population PK model developed in Chapter four was used to predict individual subject concentration-time profiles to 120 hours of enoxaparin therapy. The time spent in the therapeutic, supra-therapeutic and sub-therapeutic ranges was computed for each subject and the dosing strategies statistically compared. This study also allowed the evaluation of the results from Chapter two from a dose-exposure perspective. Chapter six of this thesis describes a survey. The aim of this survey was to gain an understanding of how dosing strategies of enoxaparin vary in four countries, investigate if clinicians are prescribing according to the Product label, and determine the methods used to dose-individualise enoxaparin. In doing so, individual hospitals in the international community will be able to compare, critique or benchmark their own strategies to peer hospitals, as well as the current literature. The publication arising from this survey would assist in the dissemination of knowledge gained from the earlier chapters of this thesis. Chapter seven is the final discussion and conclusions of the thesis along with prospects for future research.

Optimal pharmacotherapy requires an understanding of the dose-exposure (pharmacokinetics or PK) to response (pharmacodynamic or PD) relationship. Little is known about the influence of obesity on this dynamic system as PK studies in obesity have been largely descriptive rather than explanatory. This has led to a paucity of dosing guidelines for the obese, and arbitrary dose selection in the clinic. There is a need to quantify the impact of obesity on drug clearance (CL) to ensure that exposure is matched across patients of different body compositions, thereby improving therapeutic outcomes and minimising adverse events. The global aim of this thesis was to use prior published data and new clinical trial data to understand how body composition impacts upon drug CL and renal function, and to determine how clinical study design influences the identification of these relationships. Chapter 2 of this thesis determined if conventional body size descriptors that have been used to scale drug doses to body size were appropriate. In the clinical setting, a body size descriptor commonly used for determining dose requirements is total body weight (WT), based on the assumption that physiological function and PK parameters vary according to body size. However, dosing algorithms based on WT might be unsuitable for the obese due to their altered body composition which, if inaccurate, could ultimately lead to overdoses. Alternative body size descriptors such as body surface area and ideal body weight have been used, but are limited when extrapolated to obese patients as they do not take into account the covariates required to describe differences in body composition between individuals. In contrast, it was demonstrated that lean body weight (LBW), as derived by Janmahasatian et al, had the potential to scale CL across a wide range of body compositions. This literature review and systematic analysis of previously published obesity data led to the proposal of a hypothesis that body composition is sufficient to explain the influence of obesity on drug CL and that dosing for obese patients should be based on LBW. When conducting clinical studies, the selection of an appropriate body size descriptor for scaling doses across individuals of different body compositions can be aided by a study design that allows for the identification of parameter-covariate relationships which are transportable to the obese. Chapter 3 of this thesis quantified the probability of identifying these parameter-covariate relationships as a function of differing study designs. Demographics were generated using a multivariate lognormal covariate distribution with truncation at different WT limits under both a non-stratified and stratified design. PK data were simulated from a 1-compartment, first order input, first order elimination model with LBW as the covariate on CL, termed the ‘True Model’. The ‘False Model’ had WT as the covariate on CL. Both models were fitted to the simulated data and the preferred model was selected based on the difference in objective function values. Each design was evaluated under differing magnitudes of random effects, as well as under a D-optimal sparse sampling scheme. It was shown under a simulation platform that the use of stratification and a wide covariate range enhanced the probability of selecting the true covariate from two competing covariate models. The aforementioned findings regarding LBW and stratification were used to design a new clinical study investigating the influence of obesity on renal drug elimination pathways. This work forms Chapters 4 and 5 of this thesis. Non-obese and obese healthy volunteers were recruited using a study design stratified for LBW. These subjects were administered a combination of four renal markers for the simultaneous assessment of various renal processes. One of the renal markers was para-aminohippuric acid (PAH), which provides an estimation of renal plasma flow (RPF). A population PK model was developed for PAH, which revealed that body size alone was insufficient to explain variability in RPF across healthy individuals of a large range of body compositions, although LBW emerged as the preferred covariate (p=0.053) among the body size descriptors tested. This weak covariate effect was in contrast with prior research supporting the use of LBW in normalising the effect of obesity on glomerular filtration rate (GFR), implying that body composition could play a greater role in influencing GFR than RPF. This thesis has applied new methods to the design of drug CL studies in obesity, and offered results and future directions to maximise the information gained from such clinical studies. A better understanding of alterations in PK and physiological function arising from changes in body composition should aid in optimising dose adjustments for obese patients, which is of great importance given the increasing prevalence of obesity in today’s society.