Each year, around 10 million of individuals develop active tuberculosis (TB). Worldwide, TB is the leading cause of death from an infectious agent surpassing both malaria and HIV. Current treatment regimens are long and therefore encompass a risk of nonadherence and development of acquired drug-resistance, reflected in the increase of multidrug-resistant (MDR) and extensively drug-resistant (XDR) TB. Indeed, this calls for prudent use of existing TB drugs and improvement of TB treatment strategies. The aim of this thesis was to investigate the current drug susceptibility testing (DST) breakpoints for Mycobacterium tuberculosis (M. tuberculosis), the pharmacokinetics and pharmacodynamics (PK/PD) of TB treatment and to explore the role of therapeutic drug monitoring (TDM) for optimising TB treatment.

Drug resistance in M. tuberculosis is expressed over a continuous scale and for some drugs it may be identified as low- and high-level resistance. This has been poorly reflected in currently used binary susceptibility breakpoints for TB drugs. Results from genome sequencing and phenotypic DST of ofloxacin and levofloxacin were compared in study I and current breakpoints were found to misclassify up to 25% of M. tuberculosis isolates with resistance mutations in gyrA as susceptible to fluoroquinolones. This finding may have implications for the classification of XDR-TB, treatment of MDR-TB and the evaluation of fluoroquinolones in clinical studies.

Study II was a prospective cohort study of susceptible TB in Sweden, where drug concentrations of first-line TB drugs were measured along with the susceptibility level of the bacteria defined by the minimum inhibitory concentration (MIC) of M. tuberculosis. First-line drug concentrations below the reference range (16-42%) were common and most pronounced for rifampicin (13/31, 42%). An exploratory investigation of PK/PD parameters displayed a wide distribution of ratios between drug exposures and MICs. Rifampicin exhibited higher level of individual fluctuations over time during TB treatment compared with isoniazid. In study III the plasma drug concentrations of rifampicin were compared to the tuberculosis drug activity assay (TDA) and results showed that rifampicin drug levels, but not drug levels of the other first-line drugs, correlated with TDA. Patients with rifampicin drug levels below 8 mg/L had significantly lower median TDA. This finding supports the use of TDA as a potential indicator for low rifampicin exposure in resource-constrained settings without access to drug concentration analysis. The study design in study II has been further developed in study IV, which is a prospective cohort study of MDR-TB in China, where drug exposure will be explored in relation to individual bacterial MIC and measurements of treatment outcome.

In summary, the work in this thesis emphasises the importance of reliable DST of M. tuberculosis and the need to re-evaluate the currently used breakpoints. Therapeutic drug monitoring (TDM) based on drug concentrations and MICs is a useful tool to avoid suboptimal drug exposure and to individualise TB treatments. Such strategies may improve treatment regimens and avoid further development of resistance.