Should isoniazid preventive therapy follow TB treatment in HIV/TB coinfection?

A paper in the latest issue of the journal AIDS, following a presentation at last year's international conference on AIDS in Barcelona, makes a case for using isoniazid to prevent recurrence of TB in people with HIV after they are successfully treated for active tuberculosis.

The evidence takes the form of an observational study, not a randomised controlled trial, raising some questions about the strength of the findings, although the size of the study seems to have been large enough to support the main conclusions.

The researchers compared 338 HIV positive gold miners in one South African mine who received isoniazid (INH) with 221 who did not, and found there was a much lower rate (a reduction of 55%) of active TB in those who were treated with INH than among those who were not. The treatment group also received co-trimoxazole (CTX), regardless of CD4 count, whereas the "untreated" group only received CTX when their CD4 count was below 200 (or 250, if they were symptomatic). However, this did not affect the INH-related reduction in the risk of TB, which was the same regardless of CD4 count.

Death rates were not significantly reduced in the INH plus CTX group compared to the control group, although in crude terms they were lower, by 25-30%. Nonetheless, treatment provision could easily be justified based on reduced rates of serious illness.

The two groups were closely matched, except that the men who did not receive INH were generally entered in the study some months after their tuberculosis had been cured. (Matched CD4 counts suggest this was not a significant influence on their risk of TB recurrence.)

The protective effect of INH appears to be greatest in men with lower CD4 counts (below 200), where treating 5 people is enough to prevent one case of active TB. When the CD4 count is higher (above 200) it needs 17 people to be treated to prevent one case. The protective effect may disappear when individuals have experienced more than one episode of TB, although the numbers of men in this category were not large enough to be confident of this finding. There were low rates of INH resistance seen among MTb samples tested in this study, but rates would almost certainly be much higher among people repeatedly treated for TB. On theoretical grounds, it would therefore seem to be sensible to limit INH monotherapy to people who had either not suffered from active TB or had only one episode, known to have been successfully treated.

The authors do not suggest that INH treatment should always be provided to people who have been treated for TB. However, where treatment and diagnostics are good enough to prevent under-treatment of active disease, and where the risk of reinfection is high, this study indicates that this may be a valuable public health measure.