Determinants of tuberculosis Treatment Outcomes

Abstract

Background: Despite 6 months of standardised tuberculosis treatment, unsuccessful treatment outcomes (acquired drug resistance, failure, relapse) may occur in a subset of patients with drug-susceptible tuberculosis. The aim of this thesis was to study determinants of treatment outcomes in rifampicin susceptible pulmonary tuberculosis in a setting with a high prevalence of HIV-1 co-infection. Methods: A prospective cohort study was carried out to determine the frequency and determinants of acquired drug resistance. The minimum inhibitory concentration (MIC) of sputum TB isolates was determined in a sub-set (BACTEC 960 system). The clinical utility of MTBDRplus performed directly on 2-month sputum for treatment monitoring was assessed. Expression of 12 MTB-specific messenger (m) and small (s) RNAs was quantified in RNA extracted from sputum before tuberculosis treatment and compared to expression in exponential and stationary phase H37Rv cultures. Intensive pharmacokinetic studies were carried out for rifampicin, isoniazid and pyrazinamide and non-linear mixed effects modeling was used to derive individual pharmacokinetic parameters. Multivariate logistic regression and multivariate adaptive regression splines (MARS) analyses were used to evaluate utility of pharmacokinetic/pharmacodynamics measures (AUC0-24:MIC, Cmax,:MIC, % time above MIC) to predict 2-month culture conversion. A panel of analytes measured by ELISA and Luminex in plasma/serum taken pre-trearment and at 2 and 5-6 months on treatment was evaluated to determine correlates of HIV-1 co-infection, disease severity and unsuccessful outcomes. Results: 0.3- 1% of patients treated for rifampicin susceptible-tuberculosis (n=306) acquired drug resistance during or subsequent to treatment. None of the 17 cases of isoniazid monoresistance had unsuccessful outcomes. There were no differences in baseline MIC profiles (n=109) when analysed by retreatment, 2-month culture conversion or HIV-1 status and there was no evidence of increase in MIC during treatment comparing baseline and 2-month isolates (n=20). Direct MTBDRplus (n=289) on 2-month sputa had a sensitivity of 78%(95%CI 65-87) and specificity of 80%(95%CI 74-84) for predicting culture conversion with a negative predictive value of 93%(95%CI 89-96). This was in comparison to a gold standard of MTBDRplus on positive 2-month cultures. The transcriptomic signature (here limited to 11 selected mRNAs/sRNAs) of sputum (n=19) more closely resembled H37Rv in exponential phase culture than stationary phase culture. There was evidence of biological variation and a trend towards an increased hspXhi atpAlo nuoGlo signal in the sub-group who were culture positive at 2 months. Ambulant HIV-1-tuberculosis co-infected patients (n=100), the majority of whom were co-prescribed ART, did not have reduced anti-tuberculosis drug concentrations compared with HIV-1 uninfected tuberculosis patients. No relationship was found between pharmacokinetic exposures and 2-month culture conversion using logistic regression, after adjusting for MIC and clinical factors. However, MARS identified negative interactions between isoniazid Cmax and rifampicin Cmax/MIC ratio on 2-month culture conversion. If isoniazid Cmax was below 4.6 mg/L and rifampicin Cmax /MIC below 28, the isoniazid concentration had an antagonistic effect on culture conversion. For patients with isoniazid Cmax>4.6 mg/L, higher isoniazid exposures were associated with improved rates of culture conversion. We found significant differences in the majority of blood analytes over the course of treatment (n=133). Within the HIV-1 co-infected sub-group, there were marked differences comparing subjects with and without advanced immunosuppression and with and without virological suppression. We found an association between Type 1 interferon response, neutrophil-associated cytokines (IL-17 and IL-8), neutrophil-associated MMPs (MMP-8 and MMP-9) and the presence of cavitatory disease. On multivariate analyses, receiver operator curve-derived thresholds at baseline for IL-8, IL-6, IL-1RA, MMP-3 and ferritin and thresholds for IL-8. IL-6, IL-1RA, MMP-3, MMP-8, TIMP-1, TIMP-3, GM-CSF and VEGF at 2 months were able to predict unfavourable outcomes with reasonable performance characteristics. Conclusion: Although rare, in this high burden setting, acquired drug resistance does contribute to the growing drug-resistant tuberculosis epidemic. Emphasis should be on prevention of transmission to others. Ambulant HIV-1 co-infected patients, the majority of whom were co-prescribed antiretroviral therapy, did not have reduced anti-tuberculosis drug concentrations compared with HIV-1 uninfected patients. The majority of patients had plasma drug exposures well below accepted thresholds, but nevertheless had good treatment outcomes. Further clinical studies are required to investigate the potential role of nonlinear pharmacokinetic/pharmacodynamic relationships such as concentration-dependent antagonism. A deeper understanding of host and mycobacterial factors influencing treatment outcomes is required in the search for biomarkers and surrogate endpoints predicting long-term outcomes.