Efavirenz and nevirapine are the most widely used agents for treatment of HIV in children. Sources of variability in their pharmacokinetics and its association with virological outcome and adverse events in African children are poorly characterised, thereby limiting treatment optimisation. To fill this gap we studied population pharmacokinetics (PK) of efavirenz and nevirapine in 478 children from CHAPAS-3 (aged 0.3-1.5 years) using non-linear mixed effects modelling and identified predictors of treatment outcome and PK thresholds most predictive of increased risk of nonsuppression using Cox proportional hazards regression models and likelihood profiling. Efavirenz PK was described by 2-compartment disposition model with 1st-order elimination and transit-compartment absorption and nevirapine by 1-compartment model with elimination through a well-stirred liver model and transit-compartment absorption. Combined effect of SNPs 516GT and 983TC was the strongest predictor of between-subject variability in clearance (89% and 68% decrease between fastest/slowest CYP2B6 metabolic subgroups for efavirenz and nevirapine, respectively). PK was affected by weight, described with allometric scaling. Nevirapine intrinsic clearance displayed diurnal variations (oscillations of amplitude 29%, maximum at 12 noon), while age affected pre-hepatic bioavailability (31.7% lower at birth and increasing exponentially). In antiretroviral treatment (ART)-naïve children (n=325) increased exposures were associated with decreased risk of non-suppression for both drugs. In efavirenz, risk further increased for children>8 years and for younger boys; in nevirapine for high pre-ART VL, low pre-ART CD4% and low adherence. Thresholds most predictive of non-suppression in efavirenz were: Cmid-dose=1.12 mg/L, Cmin=0.65 mg/L and AUC0-24=28 mg∙h/L, while nevirapine had no clear threshold. Adverse events were infrequent in efavirenz, whereas in nevirapine transient transaminase elevations >grade 1 were associated with Cmin>12.4 mg/L. Non-nucleoside reverse transcriptase inhibitors dosage guidelines for African children should take into consideration the combined effect of SNPs CYP2B6 516G>T and 983T>C, in particular for efavirenz where we observe 10-fold differences in exposures between metabolic subgroups. Moreover the target Cmin and AUC0-24 could be lowered in children for efavirenz. Treatment initiation at lower pre-ART VL and higher pre-ART CD4%, increased adherence, and maintaining average Cmin higher than current target could improve virological suppression of African children treated with nevirapine without increasing toxicity.

Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis (M.tb). According to the World Health Organization (WHO), TB is one of the top 10 causes of death worldwide and the leading cause of death from a single infectious agent. There were an estimated 1.2 million TB deaths among HIV negative people in 2019 and an additional 208 000 deaths among HIV positive people. Although there are drugs that can cure TB, drug-resistant TB continues to be a public health threat. In 2019, there were about half a million new cases of rifampicin-resistant TB (of which 78% had multidrug-resistant TB). Novel drugs need to be isolated and developed to combat this spread of drug resistance. Historically, natural product screening has been an effective tool in anti-tuberculosis drug discovery. Examples include rifampicin, derived from rifamycin which was originally isolated from the bacterium Amycolatopsis rifamycinica and streptomycin originally isolated from Streptomyces griseus. Both these bacteria belong to the phylum Actinobacteria. Due to this history, filamentous actinobacteria were selected for screening of novel antimycobacterial compounds. However, the focus in this investigation was on species from understudied genera, the so-called “rare actinobacteria” that have not been extensively explored for antimycobacterial compounds – to increase the likelihood of discovering novel antimycobacterial compounds. The actinobacteria were assessed for antimycobacterial properties using Mycobacterium aurum A+ by the overlay method. M. aurum has been shown to be a suitable indicator for Mycobacterium tuberculosis, is safer to work with, and more amenable to high-throughput screening. Once active strains were identified by the overlay method, the zone of inhibition was excised, and the compounds were extracted using ethyl acetate. The extracts were analysed with a high-resolution mass spectrometer, and their mass spectral data were analysed using Global Natural Products Social (GNPS) molecular networking methodology. Molecular networking allowed for the dereplication of known compounds early in the drug discovery process. Strains Amycolatopsis circi S1.3T , Actinomadura strain M2 grown in ISP2 medium and Kribbella speibonae SK5 produced active fractions. These compounds could not be matched to any known compound in the GNPS database and, therefore, could be novel antimycobacterial compounds.

The epidemiology of HIV and TB are overlapping, particularly in sub-Saharan Africa, and TB infection remains common in HIV-positive children. The combined administration of anti-tubercular and antiretroviral therapies(ART) may lead to drug-drug interactions potentially needing to be addressed with the adjustment of doses. This thesis assessed the pharmacokinetics of abacavir and ethambutol and evaluated the influence of covariates such as age and concomitant medication on the PK parameters across different studies using nonlinear mixed-effects modelling. The models developed were used to estimate area under the concentration-time curve (AUC) and maximum concentrations (Cmax) achieved with the currently-recommended weight-adjusted doses. A web-based paediatric dosing tool, which is meant to be used as a first step in the design of clinical trials for paediatric dosing was also developed. The model describing the pharmacokinetics of abacavir found: a) abacavir exposure to be 18.4% larger (CI:7.50-32.2) after the first dose of ART compared to abacavir co-treated with standard lopinavir/ritonavir for over 7 days, possibly indicating that clearance is induced with time on ART, b) malnourished HIV infected children had much higher exposures but this effect waned with a half-life of 12.2 (CI: 9.87-16.8) days as children stayed on nutritional rehabilitation and recovered, c). during co-administration of rifampicin-containing antituberculosis treatment and super-boosted lopinavir/ritonavir, abacavir exposure was decreased by 29.4% (CI: 24.3-35.8), d) children receiving efavirenz had 12.1% (CI: 2.57-20.1) increased abacavir clearance compared to standard lopinavir/ritonavir. The effects did not result in abacavir exposures lower or higher than those reported in adults and are not likely to be clinically important. The ethambutol model found lower concentrations than those reported in adults. The predicted ethambutol median (IQR) Cmax was 1.66 (1.21-2.15) mg/L for children on ethambutol with or without ART (excluding super-boosted lopinavir/ritonavir) and 0.882 (0.669-1.28) mg/L for children on ethambutol with super-boosted lopinavir/ritonavir, these are below the lower limit of the recommended Cmax range of 2 mg/L. During co-administration with super-boosted lopinavir, ethambutol exposure was decreased by 32% (CI: 23.8-38.9), likely due to drug-drug interaction involving absorption, metabolism or elimination. Bioavailability was decreased in children who were administered ethambutol in a crushed form, with an estimate decrease of 30.8% at birth, and an increase of 9.6% for each year of age up to 3.2 years where bioavailability was now similar to children taking EMB full tablet. The developed paediatric dosing tool contains two major sections. a) the ‘generic module’ which uses allometric scaling -based predictions to explore the expected AUC for a generic drug, b) the ‘drug-specific module’ which can simulate entire pharmacokinetic profiles (concentration over time after dose) by using a drug-specific population pharmacokinetic model. In summary, under the current weight-adjusted doses, abacavir exposure remained within the adult recommended levels. Ethambutol dose adjustment would be required in order to achieve adult exposures. A web-based paediatric dosing tool that uses allometric scaling -based predictions as well as drug specific predictions based on published pharmacokinetic models was successfully developed.

Background: HIV remains a major contributor to the burden of disease in the Eastern and Southern African region, where around half of those with HIV/AIDS reside, according to the 2016 UNAIDS estimates. Data on the direct costs and outcomes of providing health care are important due to competing health needs and limited budgets in resource-limited settings, especially if we are to reach the UNAIDS 90-90-90 goals. This thesis presents a series of studies, which together represent the typical journey followed within an economic evaluation, starting with the establishment of a cohort, then onto cost and outcome analyses and, finally, the development of a Markov model for the purpose of establishing the cost-effectiveness of a particular intervention. Methods: Data for this thesis come from several cohorts within South Africa, with patients commencing ART between 1998 and 2014, and with care provided within a number of different models: private (Aid for AIDS), public-private partnerships or PPP (BroadReach), and public sector (Khayelitsha). The study design for all were retrospective cohort analyses. These cohorts had important strengths in their data: adherence measures (private, PPP); initiating ART at CD4 counts > 200 cells/µL (private); detailed cost data (private); long duration of follow-up with a larger proportion on second-line ART (private); ability to assess health care utilization pre-ART and in patient loss to ART follow-up (private); and availability of national identity numbers, allowing us to confirm mortality from national death register data (private, PPP). Results: The results sections of this thesis are presented in the form of published papers and chapters. In the first analysis (Chapter 4), we present a cohort profile for Aid for AIDS, where we describe the history of the programme and contrast it with the public sector programme in South Africa. In the second analysis (Chapter 5), we present a paper highlighting the profile and determinants of costs on ART over time in the private cohort. We draw attention to the impact of baseline stage and adherence to ART on early and late costs respectively. In the third analysis (Chapter 6), we explore different models of HIV care: GP versus clinic for public sector patients and courier versus collect pharmacy for private sector patients. In the third analysis (Chapter 7), we present a paper which reviews cost-effectiveness studies in LMICs and explores the relative impact of various factors on costs and mortality in preparation for the final analysis (Chapter 8), which required the development of a novel HIV Markov model. Conclusion: Interventions, such as public-private partnerships with GPs or home-refill by courier, which we have found to be associated with lower costs and improved outcomes respectively, should be considered for implementation in South Africa, especially in light of the proposed National Health Insurance. The focus of this thesis on models of ART delivery and the inclusion of under-represented or novel models are significant strengths.

Includes abstract. / Includes bibliographical references (leaves 63-66). / The aim of the study was to evaluate the impact of a school-based malaria education intervention and its effectiveness in the changing knowledge and practices related to malaria at randomly selected schools in Maputo Province.

Includes bibliographical references.

Includes bibliographical references / 4β-hydroxycholesterol (4β-OHC) is a metabolite of cholesterol formed by Cytochrome (CYP) 3A4/5/7 enzymes. It has recently been proposed as an endogenous biomarker forCYP3A4/5/7 activity. This may be useful in prediction of drug-drug interactions and other metabolic processes affected by regulators of CYP3A activity. The aim of this study was to develop and validate an LC/MS/MS assay for the determination of 4β-OHC in human plasma and use 4β-OHC as a biomarker of CYP3A4/5/7 metabolism in HIV-infected children with and without treatment in Africa. Determination of 4β-OHC from plasma was performed by saponification and derivatisation reaction processes followed by high performance liquid chromatography with MS/MS detection on an AB Sciex Qtrap 5500 mass spectrometer. Since 4β-OHC is an endogenous metabolite in human plasma, a stable isotope labelled (SIL) analogue, 4β-OHC-D7, was used as a surrogate analyte for the preparation of calibration standards and quality controls. A second SIL analogue, 4β-OHC-D4 was used as the internal standard. The transitions of the protonated derivatised products were monitored atm/z 613, 620 and 617 to the product ions m/z 490, 497 and 494 for 4β-OHC, 4β-OHC-D7and 4β-OHC-D4 respectively. The calibration curve fitted a quadratic (weighted by1/concentration2) regression over the range 2-500 ng/ml. Validation accuracy and precision statistics summary for three consecutive runs were between 98.9% and 103%, and 3.5%and 12% respectively of all quality controls. The assay's recovery, selectivity and analyte stability were established. The validated assay was successfully applied on clinical samples, where 4β-OHC was used as a biomarker to investigate the levels of CYP3A induction in HIV-infected children with and without treatment containing non-nucleoside reverse transcriptase inhibitors (NNRI).It was found that plasma 4β-OHC concentrations at baseline were significantly lower in children belonging to the naïve group compared to nevirapine (NVP) and efavirenz (EFV)groups. When NVP and EFV groups were compared at non-baseline treatment weeks, the median 4β-OHC concentrations were significantly higher in EFV group than the NVP group. Regarding the effect of time on treatment, a significant increase in 4β-OHC concentrations was observed from baseline to each of the non-baseline weeks in naïve group. Conversely, in the NVP group, there was a significant decrease in 4β-OHC concentrations from baseline to each of the non-baseline weeks. Time did not show any significant effect on 4β-OHC concentrations in EFV group. Furthermore, at baseline, age, sex and weight did not affect 4β-OHC concentrations in all the three groups. This study has provided a method that would be utilised to determine plasma 4β-OHC concentrations using relatively small volumes - typical of samples taken from children. The results of this study suggest that children on antiretroviral therapy (ART) are at risk of effects of CYP3A induction, as indicated by the increase of 4β-OHC concentrations in the NVP and EFV groups. Additionally, prolonged use of the ART may activate some nuclear receptors that regulate CYP3A enzyme activity thereby negatively affecting, for example, the regulation of lipid and glucose metabolism. The developed method may therefore be useful in predicting drug-drug interactions in the context of multiple therapy and may also be used in predicting other metabolic processes affected by regulators of CYP3A activity. Further prospective studies with larger sample sizes are required to confirm and build on the evidence shown in this study.

Introduction: Combination antiretroviral therapy (ART) has considerably reduced both the morbidity and mortality of Human Immunodeficiency Virus (HIV) infection and its associated complications, thus effectively transforming a fatal disease into a manageable chronic condition. However, the chronic use of ART has been accompanied by the emergence of adverse metabolic abnormalities in HIV-infected patients, including dysglycaemia. There is, however, a paucity of data from sub-Saharan Africa on the incidence and risk factors associated with new onset diabetes mellitus in the HIV-infected population. Furthermore, efavirenz is the preferred nonnucleoside reverse transcriptase inhibitor (NNRTI) in first-line antiretroviral therapy (ART) regimens in low- and middle-income countries, where the prevalence of diabetes is increasing. Randomized control trials have shown mild increases in plasma glucose in participants in the efavirenz arms, but no association has been reported with overt diabetes. This study explores the risk factors and incidence of diabetes, and in particular the association between efavirenz exposure and diabetes, in a large Southern African cohort commencing NNRTI-based first-line ART. Subjects and Methods: The study cohort included HIV-infected adults commencing NNRTI-based ART in a private sector HIV disease management programme from January 2002 to December 2011. Incident diabetes was identified by the initiation of diabetes treatment. Patients with prevalent diabetes were excluded. The incidence of diabetes in patients receiving efavirenz versus nevirapine containing regimens was compared with a Kaplan-Meier plot and a log-rank test. The association of efavirenz exposure with the hazard of developing diabetes was modelled using a multivariate Cox-proportional hazards model. The following variables were adjusted for in the regression model: age, sex, baseline BMI, baseline CD4, baseline viral load, exposure to diabetogenic drugs, and nucleoside reverse transcriptase inhibitor (NRTI) exposure. Results: Between January 2002 and June 2011, 62,467 patients commenced ART in the AfA program, of whom 56,298 patients met the inclusion and exclusion criteria and were included in the analysis. Median follow-up was 1.56 years (interquartile range (IQR): 0.71- 2.79 years), 21.7% of patients were followed up for 3 or more years. New onset diabetes was identified in 1500 (2.66%) patients over 113,297 patient-years of follow-up (PYFU), giving a crude incidence of 13.24 cases per 1000 PYFU. In the multivariate analysis treatment with efavirenz rather than nevirapine was associated with increased risk of developing diabetes (hazard ratio 1.27 (95% confidence interval: 1.10 - 1.46). Zidovudine and stavudine exposure, older baseline age, elevated baseline BMI, and exposure to diabetogenic medication were also associated with increased risk of diabetes. No association was found between baseline CD4 and an increased risk of diabetes. There was an association between the lowest stratum of baseline viral load and an increased relative risk for developing diabetes, but no association with higher viral load strata. Conclusion: Treatment with efavirenz, as well as stavudine and zidovudine, increased the risk of incident diabetes. Interventions to detect and prevent diabetes should be implemented in ART programmes, and use of antiretrovirals with lower risk of metabolic complications should be encouraged.

Background. Warfarin is the most commonly used anticoagulant for both primary and secondary prevention of thromboembolism. For anticoagulation efficacy, the international normalised ratio (INR) needs to be within the therapeutic range for at least 65% of time on warfarin. Objectives. To describe INR control in patients on long-term warfarin and identified predictors of good INR control at two dedicated warfarin follow-up clinics in Cape Town, South Africa (SA). Methods. We reviewed clinical records of patients in care at the INR clinics at Mitchell’s Plain Community Health Centre and Groote Schuur Hospital. We included patients who had been on warfarin therapy for at least 27 months and excluded patients with <6 months of INR monitoring data or a >70-day gap between INR tests in the calculation period, and if >25% of follow-up time was at an alternative site. The time in therapeutic range (TTR) over 180 days using the Rosendaal method was calculated, and we categorised INR control as good if the TTR was ≥65%. We constructed a multivariate logistic regression model to identify associations with good INR control. Results. We included 363 patients, with a median age of 55 years (interquartile range (IQR) 44 - 64), of whom 65.6% were women. The most common indications for warfarin were valvular heart disease (45.7%) and atrial fibrillation (25.1%). The mean TTR was 47%, with only 91/363 patients having good INR control. In a multivariate model adjusted for age, sex, clinic and target INR, patients aged ≥55 years were more likely to have good INR control than younger patients (adjusted odds ratio 1.69, 95% confidence interval 1.03 - 2.79). Poorly controlled patients had more frequent INR monitoring than those with good INR control, with a median of 8 INRs (IQR 6 - 10) v. 6 INRs (IQR 5 - 8) in the 180-day period (p<0.0001). Conclusion. Only 25.1% of patients in our study achieved good INR control, despite regular INR monitoring. There is an urgent need to improve anticoagulation control of patients receiving warfarin in SA. Validated dosing algorithms are required, and access to lower warfarin dosage formulations may optimise individual dose titration. Advocacy for these formulations is advised.

Includes bibliographical references (leaves 42-47). / Background: Nosocomial infections due to multi-drug resistant Acinetobacter baumannii are often treated with colistin, but there are few data comparing its safety and effectiveness with other antimicrobials, particularly the aminoglycosides. Metbods: A retrospective cohort study of patients treated with colistin or tobramycin for A. baumannii infections in intensive care units (ICUs) at Groote Schuur hospital was performed. Colistin was used for A baumannii isolates which were resistant to all other available antimicrobials. Tobramycin was used when the organism was susceptible to this antimicrobial. We assessed and compared ICU mortality, nephrotoxicity and time to the first negative culture in the colistin and tobramycin groups.