Nathan Geffen, TAC

Khan and colleagues published a systematic review and meta-analysis in Clinical Infectious Diseases that evaluated the impact of three factors on failure, relapse and death during treatment of TB/HIV co-infected people. These were duration of rifampicin or rifabutin treatment, (2) daily versus intermittent dosing and (3) ART during TB treatment. [1]

Out of 5,128 titles identified by a literature search, 30 articles were ultimately included in the review. These described 26 studies, of which only six were randomised trials. All reported failure and death during TB treatment and 17 reported relapse rates. Only one study examined rifabutin, so the authors considered all rifamycins together.

The authors compared the findings of three trials with internal head-to-head comparisons of rifamycin treatment strategies, but pooling the results was impossible because all three used different durations of rifampicin treatment.  In one of the trials, relapse and failure were similar in patients receiving 6- versus 9-month regimens. In another relapse rates were significantly lower after 12 months of rifampicin versus 6 months. In the third, patients who received rifampicin for 6 months of an eight month regimen had lower failure and relapse rates than those receiving it for only two months of an eight month regimen.

A pooled analysis of all 27 studies found that longer duration of rifamycin (≥ 8 months v 6 months v 2 months) was associated with less failure (1.9% v 2.6% v 2.9%), less relapse (3.3% v 9.8% v 10.8%) and less death during treatment (11.7 v 10.5 v 16.6), but the confidence intervals are very wide on all of these data and none reached significance because there were too few studies with too few participants. There were approximately 2,500 subjects in the studies that considered treatment failure, 1,300 in the studies that considered relapse and 3,000 in the studies that considered death during treatment.

However, in an analysis adjusted for a range of potentially confounding factors, rates of relapse and death were significantly higher among patients who received rifamycin for only two months compared to patients who received it for at least 8 months (relapse RR for two month rifamycin: 3.6 95%CI 1.1-11.7; death RR: 1.8 95%CI 1.0-3.1).

In studies where some patients were on ART during TB treatment there were significantly lower rates of failure (3.2% [95%CI 2.0-4.5] v 0.8% [95%CI 0-1.8]) and relapse (15.5% [95%CI 4.5-25.5] v 0.5% [95%CI 0-1.7]). But after adjustment this was non-significant.

Daily treatment was non-significantly better than three times weekly. However in adjusted analysis it resulted in significantly better relapse and failure rates during the intensive (or initial) phase of treatment (failure RR of intermittent treatment: 4.0 95%CI 1.5-10.4; relapse RR: 4.8 95%CI 1.8-12.8).

The authors describe several limitations of their analysis: there was only one randomised trial that assessed the schedule of treatment administration and only three trials that compared rifamycin duration. Because the pooled analysis mainly used observational data it was subject to confounding and selection bias. There was substantial variability in settings, patient characteristics, interventions and outcomes assessed.

comment

This study suggests the need for a clinical trial to determine the optimal duration of rifamycin during TB treatment. It also underscores the lack of compelling data on how to optimise TB treatment, contrasted with the large number of trials that have been done on ART. As the authors say, “the most important and striking finding of this review is the paucity of well-designed and adequately powered randomised trials of HIV-TB co-infection treatment.”

Reference

Khan F et al. 2010. Clin Infect Dis 2010:50 (1 May)