Background: Tacrolimus, mycophenolate and prednisolone are the current maintenance immunosuppressive agents of choice in kidney transplantation. Suboptimal dosing of these medications impacts recipient quality of life and patient and graft survival. Establishing means of tailoring immunosuppression to individual patient requirements is one of the greatest challenges facing clinicians in contemporary practice. Aims: The principal aim of this thesis was to examine the clinical utility of pharmacokinetic, pharmacogenetic and pharmacodynamic monitoring of immunosuppression in adult kidney transplantation. Specific aims included i) to establish a single multiple regression derived limited sampling strategy (LSS) capable of accurately predicting exposure to tacrolimus, mycophenolic acid (MPA) and prednisolone in our kidney transplant population, and ii) to use this LSS to examine associations between drug exposure, pharmacogenetic and pharmacodynamic data and clinical outcomes. Methods: In Chapter 1, an overview is given of the pharmacokinetics of tacrolimus, mycophenolate and prednisolone in kidney transplantation. Current dosing strategies are outlined, as is the rationale behind, and the evidence supporting, therapeutic drug monitoring of these medications. Pharmacokinetic, pharmacogenetic, and pharmacodynamic techniques are also defined and discussed. Chapter 2 details the study populations and biochemical, pharmacokinetic, clinical and statistical methodologies utilised in this thesis. The results of eight original scientific studies are presented in Chapters 3 to 10. Work in Chapters 3 to 7 involved full pharmacokinetic profiling in 20 subjects to develop and validate LSSs for tacrolimus, MPA and prednisolone. Work in Chapters 8, 9 and 10 involved limited pharmacokinetic profiling in 158 patients to investigate the pharmacokinetics, pharmacogentics and pharmacodynamics of these agents in our transplant population at the Princess Alexandra Hospital (PAH). Results: In Chapter 3, it was found that both multiple linear regression-derived and Maximum a posteriori Bayesian limited sampling methods better-predicted tacrolimus exposure (as assessed by the area under the concentration-time curve from 0 to 12 hours post-dose; AUC0-12) compared to pre-dose (C0) measurement. In Chapter 4, 12 of 25 previously published multiple regression-derived LSSs for prediction of MPA AUC0-12 in cyclosporine co-treated recipients and 1 of 53 previously published LSSs for prediction of MPA AUC0-12 in tacrolimus co-treated recipients displayed acceptable (<15%) bias and imprecision in our transplant population. Application of the LSSs most appropriate for cyclosporine co-treated patients to the tacrolimus co-treated group resulted in clinically unacceptable bias and imprecision, and vice-versa. Chapter 5 compared and contrasted the predictive power of 3 different limited sampling methods for estimation of patient MPA exposure (C0 measurement, multiple regression derived LSSs and Bayesian estimation). Correlation between MPA C0 and full AUC0-12 (AUCf) calculated using the trapezoidal rule was poor. While multiple regression derived LSSs and Bayesian forecasting methods were better able to predict full drug exposure, bias and imprecision estimates associated with these methods fell outside the level considered acceptable in clinical studies (<15%). Chapter 6 found that 10 previously published LSSs for prediction of total prednisolone exposure and 15 derived LSSs for prediction of free (unbound) prednisolone exposure performed with acceptable levels of bias and imprecision (<15%) in our transplant population. Of the free prednisolone LSSs, an equation incorporating 0.25, 2 and 4-hour post-dose concentrations showed the highest predictive power (bias 0.1%, imprecision 4.6%). Chapter 7 identified a single LSS incorporating concentration measurement at 0.5 hours (C0.5), 2 hours (C2) and 4 hours (C4) post-dose that was capable of simultaneously and accurately predicting exposure to tacrolimus, MPA and free prednisolone. In Chapter 8, high between-subject variability in dose-adjusted exposure to tacrolimus, MPA, total and free prednisolone was observed; namely 20-, 13-, 6- and 7-fold variation, respectively. Both day 4 tacrolimus and MPA exposures were independently predictive of delayed graft function [2.6 change in odds for a standard deviation (SD) increase in tacrolimus AUC0-12, p=0.02; 0.23 change in odds for a SD increase in MPA AUC0-12, p=0.02], while both day 4 MPA AUC0-12 and total prednisolone AUC0-6 were independently predictive of acute rejection [0.20 change in odds for a SD increase in MPA AUC0-12, p=0.04; 0.40 change in odds for a SD increase in total prednisolone AUC0-6, p=0.03]. A markedly higher rate of acute rejection was observed in those with lowest tertile exposure to all three drugs (50% vs. 8%, p=0.001), and lowest tertile exposure to all three drugs independently predicted acute rejection [adjusted odds ratio 34.2 (95% CI 4.1, 284.4), p=0.001]. Chapter 10 examined the influence of genetic variation in the gene encoding the pregnane X receptor (the NR1I2 gene) on tacrolimus, MPA and total and free prednisolone exposure and clinical outcomes. Median (interquartile range) dose-adjusted exposure to tacrolimus at month one post-transplantation was significantly higher in individuals carrying the NR1I2 8055T variant allele, when compared with dose-adjusted exposure in wild-type individuals [20 (14, 22) µg·h/L/mg versus 15 (9, 24) µg·h/L/mg; p=0.0007]. Using multivariate regression, NR1I2 8055T carrier status was independently predictive of higher dose-adjusted tacrolimus exposure (p=0.0005) and BK viraemia (p=0.006). No NR1I2 SNP or haplotype was significantly, independently associated with total or free prednisolone or MPA exposure. In Chapter 10, higher tacrolimus exposure was independently associated with higher day 30 concentrations of the T cell activation marker soluble CD30 (sCD30; 2.2 change in odds for a SD increase in tacrolimus AUC0-12, p=0.01; 5.5 change in odds for a SD increase in tacrolimus C0, p<0.0001). In contrast, MPA and total and free prednisolone exposures were not independently associated with sCD30 concentrations. Conclusions: High between subject variability in dose-adjusted exposure to tacrolimus, MPA and prednisolone in kidney transplant recipients provides theoretical rationale for individualised dosing. Exposure to all three drugs can be accurately and feasibly estimated using multiple regression derived LSSs. Bayesian forecasting methods provide a useful alternative for tacrolimus (not tested for MPA and prednisolone in this thesis). High variability in performance of both LSS and Bayesian methodologies highlights the importance of validating any strategy prior to applying it to an alternative population. Associations between early drug exposure and clinical outcomes suggest a potential role for modified initial dosing or early therapeutic monitoring of all three immunosuppressive agents. Pharmacogenetic monitoring via analysis of NR1I2 polymorphisms holds potential for reducing the risk of over-immunosuppression and BK viraemia and warrants further investigation. As the impact of standard immunosuppressive therapies on sCD30 concentrations is complex and poorly understood, more study is needed before consideration can be given to using sCD30 concentrations in clinical practice to guide immunosuppressant dosing. Together, these studies help identify potential opportunities for individualising immunosuppressive therapy and thus for increasing the efficacy and safety of immunosuppressant drugs. Given the exploratory nature of many of the analyses, confirmatory studies are required. If findings can be successfully replicated, prospective randomised studies should follow to see whether dosing according to the relevant techniques translates into improved clinical outcomes.