Early bactericidal activity of isoniazid in pulmonary tuberculosis. Optimization of methodology. The DATRI 008 Study Group.

Abstract

Early bactericidal activity (EBA) of antituberculosis drugs is the rate of decrease in the concentration of tubercle bacilli sputum during the initial days of therapy. The study reported here was designed to optimize the methodology for obtaining precise EBA measurements. The study compared the results with two versus five treatment days; overnight sputum collections with early morning collections; and quantitative smears for acid-fast bacilli (AFB) with quantitative cultures. Isoniazid (INH) was used as a model drug. Among 28 smear-positive patients enrolled in the study in five cities in the United States, 16 were evaluable (INH-susceptible tuberculosis [TB] and adequate sputum collections). The mean baseline bacterial load was 6.69 log10 cfu/ml (SE = 0.24). Quantitative culture of 10- or 12-h sputum collections obtained on two baseline days and treatment Day 5 was the optimal method for EBA measurement. The mean 5-d EBA was 0.21 log10 cfu/ml/d (SE = 0.03; p < 0.001) and the EBA appeared to be constant during the first five treatment days. On the basis of these data, multiarm studies of investigational drugs will require 25 evaluable subjects per arm to detect (80% power and two-tailed alpha of 0.05) an EBA at least 50% as large as the EBA of INH. In countries with a low incidence of TB, the usefulness of this methodology for rapidly assessing new antituberculosis agents may be limited by the relatively large number of subjects required to compare EBA values across treatment arms.