© 2003 American Thoracic Society
The Hunt for the Elusive Surrogate Marker of Sterilizing Activity in Tuberculosis TreatmentDenver Public Health and Division of Infectious Diseases, Department of Medicine University of Colorado Health Sciences Center Denver, Colorado In this issue of AJRCCM (pp. 1348–1354), Jindani and colleagues (1) present an important analysis from a study originally published 22 years ago (2). In the rapidly changing world of clinical research, what could possibly make data from an old study of old drugs of interest today? Tuberculosis drug development had been essentially dormant for nearly 30 years, but is reawakening. Rates of resistance to the first-line drugs are relatively high in many parts of the world (3), and drug resistance decreases the effectiveness of directly observed regimens (4). Despite the success of current regimens for drug-susceptible tuberculosis, it is challenging to retain patients in care for the minimum duration of "short-course" regimens (6 months). With renewed attention, there is growing recognition of the challenging aspects of tuberculosis drug development. Tuberculosis requires multidrug therapy, but the use of multiple effective agents can make it difficult to discern the effect of a new drug. Phase 3 efficacy studies require 700 to 1,000 patients followed for at least 18 months and cost approximately 20 million dollars (5). Because a new drug can be used in many ways (different doses or dosing frequencies, different companion drugs), it is critical to have as much data as possible before making the final choice to be evaluated in a large phase 3 study. What can clinical trial designers use to make these decisions of drug, dose, and place in a multidrug regimen? Standard in vitro measures of drug activity do not correlate well with results of clinical trials (6). With careful attention to use of doses that mimic human pharmacokinetics, the mouse model correlates well with results of clinical trials (7) and is a mainstay of tuberculosis drug development. There is still a need, however, to evaluate drug activity in humans and do so in studies that do not require 1,000 patients and years of follow-up. There are two distinct phases of tuberculosis treatment (8). In the first few weeks of therapy, there is rapid killing of metabolically active bacilli. Thereafter, therapy is continued for months to prevent a relapse of the active infection after treatment discontinuation. The ability to prevent relapse is termed sterilizing activity because it is presumed to require killing nearly all bacilli remaining after the initial phase of therapy. Currently available drugs have very different degrees of activity during these two phases of treatment. Isoniazid has remarkable killing activity during the early part of therapy but has little sterilizing activity (9); pyrazinamide is exactly opposite, having little activity early in therapy but potent sterilizing activity (2, 10). A major challenge in tuberculosis drug development is finding a way to identify sterilizing activity without undertaking a phase 3 clinical trial. The only validated surrogate marker for sterilizing activity is the rate of conversion of sputum culture to negative after 2 months of treatment (11). Although valuable, 2-month culture conversion is not an ideal surrogate marker of sterilizing activity. Such studies require sample sizes of 100 to 150 patients per study arm. Therefore, there is great interest in finding a surrogate marker of sterilizing activity that requires less time and fewer patients than studies using 2-month culture conversion. The assessment of early bactericidal activity (EBA) has been suggested as a possible surrogate marker for sterilizing activity. The technique assesses changes in quantitative sputum culture over the first few days of treatment. The use of a continuous outcome measure—quantitative sputum culture—allows differences to be detected with much smaller sample sizes than are necessary for studies having a dichotomous outcome (e.g., culture-positive vs. culture-negative). Instead of hundreds of patients per study arm, EBA studies use 4 to 10 patients per study arm (1, 12). EBA studies include the many possible effects of the human pharmacokinetic behavior of a new drug (absorption rate, binding to plasma proteins, metabolism, etc.). EBA studies can use single-drug therapy for a short period of time, thereby allowing the assessment of a new drug without having to account for the effects of the other drugs in a multidrug regimen. In sum, EBA studies have many attractive features, but do they detect sterilizing activity? The initial analysis by Jindani and colleagues focused on the assessment of EBA during the first 2 days of therapy (2). These data have now been reanalyzed to more fully characterize drug activity during Days 2 to 14 (1). There are methodological controversies with EBA studies, particularly involving details of the mathematical techniques used in their analysis (1, 12), but the focus of this reanalysis is whether EBA activity after the first 2 days ("extended EBA") predicts sterilizing activity. Why re-evaluate the EBA of old drugs (isoniazid, rifampin, pyrazinamide, ethambutol, and streptomycin) whose roles in therapy have been well established in previous clinical trials? If activity in EBA studies predicted the clinical activity of these well-characterized drugs, this would give one confidence that this technique would be useful in choosing new drugs, doses, and drug combinations. Unfortunately, the results of this new analysis suggest that activity in an extended EBA study does not reliably identify sterilizing activity. Extended EBA identifies rifampin as a drug with potent activity after 2 days. It is concerning, however, that the effect of rifampin was not statistically significant when the analysis was restricted to the first 8 days (the likely maximum duration of monotherapy likely to be judged ethical now). Furthermore, other results from this study show the problems of predicting sterilizing activity from EBA. The potent sterilizing activity of pyrazinamide was not detected. Streptomycin appears to have potent activity in extended EBA and yet has little sterilizing activity in randomized clinical trials (13). Ethambutol appears to be antagonistic in extended EBA, but there is no clinical evidence that ethambutol interferes with the sterilizing effect of rifampin or pyrazinamide (14). If extended EBA does not correlate well with sterilizing activity, what then is the role of EBA studies in tuberculosis drug development? The most important uses of an EBA study may be to compare different members of the same class of drugs (e.g., the fluoroquinolones) and to compare different doses of the same drug. EBA studies do not appear to be helpful in evaluating the effect of drug combinations, but fortunately, the mouse model is very useful in this regard (7). For now, studies to assess sterilizing activity will have to continue to use 2-month culture conversion as an endpoint. The search for an efficient surrogate marker for sterilizing activity goes on. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
