© 2004 American Thoracic Society
Side Effects of Antituberculosis TherapyTo the Editor:In an article by Yee and colleagues (1), the authors found a high incidence of major side effects for pyrazinamide. Although this might be in line with clinical impression, there could have been a major methodologic pitfall. Individual susceptibility is a key factor for drug reactions. Patients tolerating the drugs initially are much less likely to develop side effects afterward. Indeed, in the article (1), 98% of major side effects occurred within the first 2 months of drug administration. Similarly, in a previous controlled trial of 2, 4, and 6 months of pyrazinamide in short-course regimens, 83% of side effects of any kind occurred within the first 2 months (2). As pyrazinamide is regularly given for 2 months, whereas isoniazid and rifampin are given for 6 months, the use of incidence rate per person-month of treatment would have introduced up to a threefold bias against pyrazinamide. Consider the following hypothetical situation: the three drugs produced an equal number of major side effects within the first 2 months, and no major side effects occurred afterward among those not experiencing side effects initially. The calculated relative risk for pyrazinamide versus isoniazid or rifampin would be 3 by person-month of treatment, but only 1 by intention-to-treat analysis on per-person basis. The Cox analysis cannot eliminate the systematic bias that was introduced. The complex statistical calculation is no more robust than a simple comparison among groups, as consistently used in all the randomized controlled trials quoted by the authors (3–5). Indeed, the nature of the data might not have allowed the authors to draw any firm conclusion about drug-induced hepatitis. All the quoted hepatitis events occurred when the drugs were used together, and two or more drugs were subsequently withdrawn, making it difficult to pinpoint the offending agent. It is also possible that the drug combination, rather than pyrazinamide alone, caused the hepatitis. As patients' safety often precludes actual rechallenge, one can never exclude the possibility that all the original drugs might be successfully put back as reported previously (6). Simple apportioning of the blame, although intuitively appealing, might not be most appropriate. The currently used 6-month short-course regimen has been established through meticulous research. However, much remains unknown about pyrazinamide, including its optimal dosage and frequency of administration among different age groups. Carefully designed research is necessary to address these critical issues for the benefit of the vast number worldwide of patients with tuberculosis.
a Department of Health Hong Kong, China FOOTNOTES Conflict of Interest Statement: C.C.L., C.K.C., and W.W.Y. have no declared conflict of interest. REFERENCES
Conflict of Interest Statement: D.P.Y. and D.M. do not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
From the Authors: We thank Dr. Leung and colleagues for their comments regarding our study (1). We used the incidence density method to account for the variable duration of exposure of pyrazinamide, isoniazid, and rifampin. This could have exaggerated differences among drugs, if adverse events occurred only in the first 2 months; but in one large-scale trial, isoniazid hepatoxicity continued to occur in the seventh to twelfth month of therapy (2). Given this, person-time estimates will be more accurate and particularly useful for comparison of risks, benefits, and costs of regimens of different duration, such as 2 months of rifampin–pyrazinamide versus 9 months of isoniazid for latent tuberculosis (TB) infection. Attribution of adverse events to individual drugs can be difficult in an observational study. However, in our study, pyrazinamide was the most commonly responsible drug for hepatitis and overall adverse events when a single drug was identified unequivocally in 36 out of 46 serious adverse events. Even if all the events of hepatitis that we divided between isoniazid and pyrazinamide had been attributed entirely to isoniazid, the incidence of isoniazid- and pyrazinamide-related hepatotoxicity would have been equal, and overall adverse events in pyrazinamide would have been twice as high as any other drug (1). We agree with Dr. Leung that carefully conducted randomized trials have established the use of pyrazinamide for therapy of active TB, but we do not agree that these trials accurately estimated the incidence of adverse events. In randomized trials of therapy for tuberculosis infection and disease, incidence of adverse events has been considerably less than observed later in routine clinical practice (3). In initial randomized, placebo-controlled trials involving more than 100,000 individuals, isoniazid hepatotoxicity caused by isoniazid was uncommon (0.1–0.3%), and no deaths were reported (4). However, after widespread use in routine practice, isoniazid hepatotoxicity was 10 times higher with associated mortality (5). Similarly, in randomized trials, occurrence of adverse effects with 2 months of rifampin–pyrazinamide was similar to that with 6 to 12 months of isoniazid; but after more widespread use, much higher rates of serious and even fatal hepatotoxicity were reported (6). The differences between the incidence of adverse events in randomized trials and routine clinical practice remain unexplained. However, they underscore the importance of careful surveillance to estimate adverse events from new regimens whose utility has been demonstrated in randomized trials, after their introduction into routine clinical use.
McGill University Montreal, Quebec, Canada REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
