© 2003 American Thoracic Society
Role of Isoniazid in Once-Weekly Rifapentine Treatment of Pulmonary TuberculosisDepartment of Medical Microbiology St George's Hospital Medical School London, United Kingdom In the current issue of AJRCCM (pp. 1341–1347), a Tuberculosis Trials Consortium report relates plasma concentrations of isoniazid and rifapentine to the outcomes of their study 22 on once-weekly treatment with these drugs (1). They provide clear evidence that the areas under the plasma isoniazid concentration–time curve were smaller and isoniazid acetylation was more rapid in patients who failed during treatment or had a relapse after it. In contrast to these findings, no association was found between rates of isoniazid acetylation and failure/relapse in an earlier Hong Kong study, and the response was similar in patients with organisms initially sensitive or resistant to isoniazid (2). The two studies had similar test regimens with a 2-month initial four-drug phase followed by 4 months of continuation therapy with once-weekly doses of 600 mg rifapentine plus 15 mg/kg isoniazid. The only difference was initial streptomycin in the Hong Kong study rather than ethambutol in study 22. Why did isoniazid influence the outcomes in study 22 but not in the Hong Kong study? Extended studies of early bactericidal activity (3) and analysis of the effects of initial isoniazid resistance in clinical trials (4) show that specific drugs predominate in the overall bactericidal action of regimens. During the first 2 days, isoniazid is the predominant drug. Its high activity is not altered by the presence of other drugs, including rifamycins, in the regimen. However, after 2 days, rifampin is predominant (acting with pyrazinamide in the initial 2-month phase only) in standard short-course regimens whereas the presence of isoniazid in the regimen has no effect on overall bactericidal activity. We first have to see how rifamycins and isoniazid act in vitro to understand this. Rifamycins are more bactericidal than isoniazid only when spurts of bacillary metabolism occur (5). Whenever metabolism starts, they bind to RNA polymerase and start to kill within seconds whereas isoniazid inhibits cell wall growth (6), and may take hours to start (5). After actively growing bacilli in lesions have been killed by isoniazid during the first 2 days, similar spurts of metabolism in the remaining persistent bacilli may well occur. Rifamycins then kill long before isoniazid can have any effect, so accounting for their predominance after 2 days. We now speculate on drug action during treatment, remembering that rifapentine is 98% bound to plasma protein (7). At the start of treatment of culture-positive pulmonary tuberculosis, the vast majority of bacilli are in cavities where they lie extracellularly in liquefying caseum (8). As inflammation in and around the cavities dies down at the end of the initial phase, blood capillaries and "lakes" near the capsular borders of the cavities regain their impermeability to plasma protein. The diffusion of rifapentine to the bacilli is greatly reduced. Some of the bacilli (we do not know how many) may have been phagocytosed. Rifapentine is about 26 times less active inside macrophages than outside them, again probably due to binding (9). Thus there is a great reduction in the availability of rifapentine both for extracellular and for intracellular bacilli. Isoniazid, however, is not plasma bound and permeates freely. Let us now look at bacillary metabolism in lesions. There must be some metabolism in persisting bacilli to allow rifamycins and isoniazid to kill at all, even if slowly. We can then consider small healing lesions, termed type A, in which the bacilli are usually dormant but show occasional spurts of metabolism. There is no cell division and so isoniazid has no bactericidal activity. Whereas a full pulse of rifapentine followed by its postantibiotic lag of about 3 days occupies almost the full week between doses, the period occupied by the available small pulse of rifapentine and its shorter postantibiotic lag of only 1 day would be greatly reduced, allowing several days free of the effect of rifapentine (10). In this drug-free period, growth but not metabolism would be largely inhibited by normal immunity either in cells near extracellular bacilli or in macrophages containing intracellular bacilli. In the four human immunodeficiency virus–seropositive patients in study 22, low immunity would allow growth and the subsequent development of rifamycin mono-resistance (11). The absence of isoniazid resistance in these relapse cultures proves that isoniazid cannot be having bactericidal activity. So why was there a prognostic effect of isoniazid in study 22? There might also be larger residual, open, cavitary lesions, termed type B. In these, inflammatory cells are blocked from reaching the residual bacterial populations within the cavity by the cavity wall (10). Their bacilli continue to multiply between small rifapentine pulses; cell division occurs, so that isoniazid is now the predominant drug and has greater bactericidal activity than rifapentine. The severity of disease and the frequency of cavitation was less in the Hong Kong patients than in study 22 (1). Type A lesions would be frequent in Hong Kong and no association with isoniazid activity would be found. In contrast, failure, common in study 22 but not in Hong Kong, and relapse would be likely to arise from type B lesions (12), so that even a single such lesion could account for the correlations between isoniazid activity and response in study 22. Furthermore, the use of streptomycin rather than ethambutol in the initial phase of the Hong Kong study would increase sterilizing activity and prevent type B lesions, as streptomycin has been shown to have early sterilizing activity as great as that of rifampin, and ethambutol may antagonize the activity of the other drugs (3). The crucial difference between the results in the two studies probably depends on the occurrence in study 22, but not in Hong Kong, of residual cavities containing persister bacilli that continue to divide. The opportunity for division arises between doses of the small pulses of rifapentine available in lesions. Division could be prevented by increasing the dose size perhaps to 900 or 1,200 mg weekly, or by the addition of another once-weekly drug, such as moxifloxacin (13) provided there was convincing evidence of its sterilising ability. In summary, the key issue distinguishing the two studies seems to be whether bacillary multiplication, not just persistent metabolism, occurs in residual cavities present in study 22 but not in the Hong Kong study. Multiplication, after small pulses of rifapentine, would allow isoniazid to kill. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
