Response to: Harrison TS, Macallan DC, Rayner CFJ, Wansborough-Jones M. Treatment of tuberculosis in HIV-infected individuals AIDS 2002

Abstract

We appreciate the opportunity to respond to the comments of Harrison et al. [1]. Whereas we fully acknowledge the criticisms regarding observational studies, we may have to agree with Voltaire in that the ‘best is the enemy of the good'. Waiting for a double-blind randomized controlled trial of highly active antiretroviral therapy (HAART) versus no HAART in patients with a CD4 cell count of less than 100 might be considered inappropriate. We agree that it would be interesting to have further divided the group who received HAART into those who started therapy within 2 months, those who started after the induction phase, and those already on HAART, but the numbers would have been too small for any meaningful analysis. It was for this reason that we included, in our initial submission, data on the timing of opportunistic infection and mortality with respect to tuberculosis (TB) diagnosis and antiretroviral therapy. As can be seen from Table 1, whereas eight out of 24 patients not receiving HAART developed their opportunistic infection before 2 months, two-thirds occurred soon afterwards, the majority being related to CD4 cell thresholds. It is not unreasonable to assume, therefore, that the early initiation of HAART may have helped prevent these outcomes.Table 1: Characteristics of HIV-positive patients with tuberculosis infection who subsequently developed further AIDS-defining illnesses. Of the 16 deaths, four did not receive TB treatment, two died within 2 months, and the remaining 10 died between 2 and 8 months. Once again, the majority of these might have been avoided with the early initiation of HAART. Harrison et al. [1] shared our concern about the prescribing of antiretroviral agents in the first 2 months of TB therapy because of the high rate of adverse drug events, many requiring the interruption of therapy (34%). Whereas comparisons with observational cohorts [2,3] may be difficult, it is interesting to note that similar high rates of modification or discontinuation of HAART have been noted. Furthermore, although many patients modified their antiretroviral regimen, the discontinuation rate in our study was very low. We hope our study has been valuable in highlighting ways in which one can attempt to lower the rates of adverse events and minimize the interruption of therapy by measures such as avoiding neurotoxic antiretroviral therapy, using antiretroviral therapy with a lower incidence of rash and hepatitis, and anticipating gastrointestinal problems. There are no published data to support the statement that immune reconstitution may be more frequent or severe in TB if HAART is started earlier. Immune reconstitution can occur if HAART is started at the initiation of TB therapy or any time after. As far as affecting adherence, the majority of our patients are given supervised or directly observed therapy as recommended in national guidelines [4]. The use of once a day regimens and a low pill burden has made this more practicable. There are no data to suggest that treatment of both diseases together lead to an increase in the TB relapse rate. Indeed our TB relapse rate is comparable to that seen in HIV-negative cohorts. The virological response to initiating HAART seen in our cohort of HIV/TB co-infected patients has been shown to be comparable to HIV-positive patients without TB [5]. We disagree with the statement that drug interactions are a reason to avoid antiretroviral therapy in the first 2 months of TB treatment, as the majority of drug interactions are seen with rifamycins and isoniazid, which are the backbone for the entire duration of TB therapy [4,6]. We find it remarkable that there is doubt that HAART given to patients with advanced HIV disease may not be associated with a reduction in mortality because Cozzi Leppri et al. [7] demonstrated a sixfold reduction in the risk of AIDS or death from 0.19/person-year to 0.03/person-year after the commencement of HAART in patients with a CD4 cell count of less than 50. The fact that it takes time for HAART to to restore the CD4 cell count and that we have shown that the majority of both opportunistic infections and death occurred after 2 months inform the debate that the early initiation of HAART is indicated in advanced disease. Our observational study will hopefully give some guidance to physicians on prescribing HAART in TB infection and highlight the problems of drug–drug interactions, adverse events and immune reconstitution. We do not think it is justifiable to leave patients with very low CD4 cell counts for a prolonged period of time without antiretroviral therapy. For patients who have an AIDS diagnosis, it is important to try and improve their immune system as rapidly as possible, in spite of the potential for drug interactions and immune reconstitution. Simon Edwardsa Anton L. Pozniakb Rob Millera on behalf of all the authors