Stan Houston and colleagues correctly point out the discrepancy between current Canadian1 and US2 guidelines, with respect to HIV- infected immigrants from high-burden countries. Earlier US guidelines included suggestions that anergic HIV-infected persons be considered potential candidates for treatment of latent TB if they belonged to groups where the expected prevalence of tuberculous infection was 10% or greater.3 More recent research has cast considerable doubt on the use of anergy testing among HIV-infected people.4
My purpose in reproducing the Canadian guidelines was to highlight them, rather than to evaluate them. The treatment of latent tuberculosis infection among anergic HIV-infected persons has not been supported by recent investigations — despite the biologic rationale when the prevalence of undiagnosed latent TB is expected to be high. For example, an estimated incidence of 3.0 cases of active TB per 100-person years was found in a cohort of anergic Italian patients with HIV infection — largely injection-drug users.5
Several of the articles cited by Houston and colleagues have limited statistical power. In a US multicentre placebo-controlled clinical trial of isoniazid among HIV-infected people with anergy and TB risk factors, substantially fewer cases of active TB occurred than expected in the placebo arm (6 cases in 257 subjects, or 0.9 per person-year), making a significant treatment effect virtually impossible to detect.6 This may have reflected concomitant antiretroviral therapy. The Ugandan study cited did not reach target recruitment in the anergic subgroup.7 Both studies yielded effect estimates compatible with some benefit of isoniazid (though less than for tuberculin-positive cohorts), but very wide confidence intervals.
These findings are consistent with anergic HIV-infected cohorts including people with and without latent TB. The risk of subsequent active TB must reflect the background prevalence of latent infection, the use of antiretroviral therapy, the impact of competing risks, and the risk of subsequent tuberculous infection. The prevalence of unidentified latent TB among untreated anergic HIV-infected individuals may be lower than that of the general population, because many would have already developed active TB related to a previous latent infection. I suspect that many practitioners do not ordinarily prescribe isoniazid in this context, although they may occasionally find good reason to do so (for example, a particularly high likelihood of previous exposure, without documentation of specific contacts).
I agree that my statement that the 2-month rifampin–pyrazinamide regimen is “clearly contraindicated for anyone with underlying liver disease or with isoniazid-related hepatotoxicity” was too strong.8 This reflected considerable concern about recent reports of liver toxicity,9 perhaps related to suboptimal patient selection and supervision, as noted by Houston and colleagues. The key point is that rifampin–pyrazinamide should not ordinarily be prescribed for latent TB when there is substantial risk of hepatotoxicity, except with expert supervision and close monitoring.
I would be particularly hesitant to use this regimen where there is evidence of active ongoing liver disease (as opposed to positive viral serologies or a history of alcohol use). Pyrazinamide is the most likely culprit when liver toxicity does occur.10,11 Hence the 4-month rifampin regimen provides a reasonable “short-course” alternative to isoniazid, where feasible.
Kevin J. Swartzman McGill Respiratory Epidemiology Unit Maison Lady Meredith House Montreal, Que.