It has generally been held that the repeated emergence of resistance in Mycobacterium tuberculosis is due to the effects of large population sizes, slow replication, and prolonged colonization and treatment. However, there have been suggestions that its emergence is facilitated by high mutation rates due to a lack of mismatch repair, error-prone polymerases, and a potentially mutagenic host niche. Genome re-sequencing has indicated higher variability in strains with emergent resistance, but these studies have not been performed in serial isolates in which drug resistance has emerged. We have used genome re-sequencing to address the mutational processes that occur during the evolution of drug resistance during a clinical infection.

Serial isolates from a patient obtained over a 12 month period, and spanning the transition of the colonizing population from fully drug sensitive, to isoniazid resistant, to isoniazid and rifampicin (multiply drug) resistant, spanning an estimated minimum of 100 generations within the host, were deep sequenced using Illumina sequencing. The genomes were compared, and all mutations in non-repetitive sequences were identified.

Specific mutations conferring resistance were identified. No additional mutations in non-repetitive regions were present. The mutations observed were kat S315T and rpoB D516Y.

M. tuberculosis is relatively stable genetically within the host, and demonstrates greater stability than is suggested by in vitro studies of emergent drug resistance, or by models of hypermutability. This indicates that it is primarily the nature and duration of the infection that are sufficient to lead to the repeated emergence of drug resistance in this infection if improperly managed, and that the selective pressure of the drugs limits additional diversification. This emphasizes the central importance of maintaining therapeutic concentrations of at least two effective antibiotics for the duration of treatment to prevent the emergence of resistance.