Pharmacokinetics of Rifampicin

TO THE EDITOR —Nijland et al. [1] present interesting data regarding the pharmacokinetics of rifampicin in patients with tuberculosis and diabetes mellitus. However, their conclusions are weakened by 2 important flaws in the study design. First, they used the area under the curve between 0 and 6 h (AUC_{0–6 h}) as a surrogate for total rifampicin exposure. Time points beyond 6 h are routinely used to examine rifampicin pharmacokinetics, because the maximum concentration may occur as late as 8 h after administration of the dose [2]. The data presented in table 2 and figure 1 of the article strongly suggest delayed rifampicin absorption in the diabetic subjects, compared with the nondiabetic subjects. The median time to the maximum concentration was 4 h (the upper limit of the range was 6 h) in the diabetic subjects, compared with 2 h (upper limit of range, 4 h) in nondiabetic subjects; even though this difference was not statistically significant, it appears to have been a significant confounder. Furthermore, the relative decrease in rifampicin plasma concentrations between 4 and 6 h was ∼47% in the nondiabetic subjects versus ∼30% in the diabetic subjects, suggesting ongoing absorption in at least some subjects during this time. Examination of the plasma rifampicin concentrations at a later point would have greatly enhanced confidence that the difference in rifampicin pharmacokinetics between diabetic and nondiabetic patients with tuberculosis was an effect of malabsorption, as opposed to delayed absorption.

Second, the authors did not match the diabetic and nondiabetic subjects by weight and sex, which have both been associated with significant differences in rifampicin exposure [2]. Linear regression is inadequate to adjust for the difference in weights among the diabetic and nondiabetic subjects, particularly given the small sample size. As can be seen in figure 2 of Nijland et al. [1], 11 of the 17 nondiabetic subjects weighed <50 kg (compared with 4 of 17 diabetic subjects), and there is a cluster of 8 nondiabetic subjects who weighed <50 kg and who had very high rifampicin AUC_{0–6 h} values, suggesting a nonlinear relationship between subject weight and rifampicin exposure at higher doses (in mg/kg). The area under the curve for nondiabetic subjects who weighed ≥50 kg falls well within the range for diabetic subjects in this study.

In short, the data presented by Nijland et al. [1] are tantalizing, but because of serious limitations, do not clearly demonstrate that type 2 diabetes is associated with decreased exposure to rifampicin.

• © 2007 by the Infectious Diseases Society of America