In order to effectively optimize and evaluate new potentially treatment-shortening strategies for tuberculosis, there is an urgent need to develop reliable surrogate end points of the sterilizing activity of antituberculosis drugs in vivo. Serial sputum colony counting can be used to directly quantify changes in sputum bacillary load during treatment and, with appropriate statistical modelling, can demonstrate the distinct phases of pharmacodynamic response which may reflect effects on putative subpopulations of Mycobacterium tuberculosis. These characteristics may allow this technique to efficiently compare the sterilizing activity of different multidrug regimens during the intensive phase of therapy, as suggested by existing studies in diverse settings. This methodology could be used to select the best drug doses and combinations to take forward into phase III clinical trials.
