Thesis (PhD (Biomedical Sciences. Molecular Biology and Human Genetics))--University of Stellenbosch, 2007. / One of the aims of direct observed therapy strategy implemented by the World Health Organization was to prevent the development of drug resistant tuberculosis. However, in recent years a dramatic increase and spread in multidrug resistant tuberculosis has been observed. In this study, a molecular epidemiological approach was used to understand and rapidly detect drug resistance in high incidence tuberculosis communities of the Western Cape, South Africa. Previous studies showed that, drug resistant tuberculosis occurs as a result of spontaneous mutations in particular genes. Using molecular techniques, we developed an algorithm to rapidly detect isoniazid, rifampicin and ethambutol drug resistance in tuberculosis patients from a short term mini culture. Rapid detection of drug resistance is important to prevent future transmission events. In addition, accurate ethambutol resistance testing is of particular importance, since treatment of patients infected with multidrug resistant strains with second line anti-tuberculosis drugs depend on the ethambutol test results. In a comprehensive study, we found that the algorithm performs well when compared to the traditional culture method currently used by the routine laboratories. However, the results showed that more then 90 % of ethambutol resistance is missed by the routine laboratories. This has important implications for the tuberculosis control program, since patients infected with the drug resistant strain may be on inappropriate treatment. In this study, we found that certain strains have a selective advantage to become drug resistant and transmit and this implies that they are more virulent and fit than other strains. This observation has also been made for strains within the same genotype family. The more transmissible drug resistant strains cause large drug resistant outbreaks. This study highlights the complexity of the drug resistant epidemic, and confirms that it is a major problem in local communities. Application of molecular methods has provided us with tools to study how resistance might develop. We have demonstrated how we made use of a newly developed method to detect a multidrug resistant outbreak in the study community. The applications of transcriptomics identified several genes that might play a role in isoniazid resistance. Using this data a model was proposed whereby isoniazid resistant strains can compensate for the toxic effect of the drug. Application of comparative genomics by whole genome sequencing will be used to assist us in the further understanding of the mechanisms of drug resistance. This study also conclude that we should continue in our attempts to develop faster diagnostics for both first and second line drugs and that we must not loose site that all of this research must in the end benefit the patients.