The trusted source for
healthcare information and
By Richard R. Watkins, MD, MS, FACP
Division of Infectious Diseases, Akron General Medical Center, Akron, OH; Associate Professor of Internal Medicine, Northeast Ohio Medical University, Rootstown, OH.
Dr. Watkins reports no financial relationships in this field of study
SYNOPSIS: A meta-analysis that included 53 studies found treatment regimens for latent tuberculosis containing rifamycins
had improved efficacy and lower toxicity compared to isoniazid.
SOURCE: Stagg HR, et al. Treatment of latent tuberculosis infection: A network meta-analysis. Ann Intern Med 2014;161:419-428.
Latent tuberculosis infection (LTBI) is a common reason for referral to infectious disease specialists. For the past few decades the standard therapy in the United States has been isoniazid for 9 months. While generally safe and effective, many patients are unable or unwilling to comply with this lengthy regimen. Thus, it is necessary to offer effective and low toxic regimens for the shortest time available to optimize compliance rates. Stagg and colleagues aimed to provide an up-to-date recommendation based on randomized, controlled clinical trials for the relative efficacies and adverse event profiles of different LTBI treatment regimens.
The study was a meta-analysis of randomized, controlled clinical trials that used a network approach, which the authors stated is a better method than conventional meta-analyses because it allows indirect comparisons of regimens and produces better inferences of relative efficiency. The two main endpoints were hepatotoxicity and development of active TB. Regimens were grouped as follows: all rifampicin regimens; isoniazid regimens 3 to 4 months in duration, 6 months in duration, 9 months in duration, or 12 months or more in duration; isoniazid-rifampicin regimens 3 to 4 months in duration; all rifampicin plus pyrazinamide regimens; and all rifampicin-isoniazid-pyrazinamide regimens. The main adverse event of interest was hepatotoxicity of grade 3 or higher, but all types and severity of adverse events were recorded. The quality of the studies was determined by using the Cochrane Collaboration’s tool for evaluating study bias.
Of 1,516 studies identified, 53 with 133,992 subjects met the inclusion criteria for the network meta-analysis. All of the tested regimens except isoniazid-rifampicin had a favorable odds ratio (OR) <1.0 compared to no treatment. Six regimens were efficacious compared to placebo: isoniazid for 6 months (OR 0.64; 95% confidence interval [CI], 0.48 to 0.83), isoniazid for 12 months or longer (OR 0.52; CI, 0.41 to 0.66), rifampicin for 3 to 4 months (OR 0.41; CI, 0.18 to 0.86), rifampicin-isoniazid for 3-4 months (OR 0.52; CI, 0.34 to 0.79), pyrazinamide-isoniazid-rifampicin (OR 0.34; CI, 0.18 to 0.62), and rifampin-pyrazinamide (OR 0.55; CI 0.33 to 0.92). The rifabutin-isoniazid regimen had a wide confidence that crossed 1, as did isoniazid for 9 months. Rifampicin-only and rifapentine-isoniazid regimens had lower rates of hepatotoxicity than isoniazid-only for 6 months or longer. Furthermore, rifampicin-isoniazid regimens had lower hepatotoxicity versus isoniazid-only therapy when the latter was given for 12 to 72 months. There was also good evidence that regimens containing pyrazinamide had higher rates of hepatotoxicity. Immunosuppression, HIV status and TB incidence did not affect the incidence of hepatotoxicity. Other serious adverse events were rare across all the studies. Rifampicin-pyrazinamide regimens had the highest risk for gastrointestinal adverse events, while rifampicin containing regimens had the highest risk for central nervous system adverse events. Five toxicity-related deaths were reported which were all due to severe hepatitis from isoniazid.
A large body of evidence has accumulated over the last few decades for treating LTBI. The present study critically analyzed these data and reported two main findings:
(1) 3 to 4 months of rifampicin monotherapy was effective in preventing the development of active TB and had a low incidence of hepatotoxicity compared to isoniazid, and;
(2) pyrazinamide, while effective, has an unacceptably high risk-to-benefit ratio.
Thus, rifampicin seems to have the optimal balance of effectiveness and tolerability. While important, neither of the two results is novel or unexpected as previous studies have come to the same conclusions. Indeed, the clear benefits associated with rifampicin monotherapy raises the question of why 9 months of isoniazid therapy remains the standard of care for LTBI in North America. Compliance with this regimen is fair at best and isoniazid-associated hepatotoxicity, while rare, can be fatal. Shorter course therapy also has the potential advantage of using fewer health care resources and perhaps lowering costs. Further economic analyses comparing standard therapy (i.e. isoniazid) versus shorter course rifamycin therapy will help to clarify this issue. Moreover, results of an ongoing trial comparing the efficacy of 4 months of rifampicin to 9 months of isoniazid and another evaluating 3 months of self-administered isoniazid-rifapentine will add to the available evidence.
One frequent criticism of meta-analyses is that they can make comparisons between studies that are too dissimilar to be meaningful, e.g. apples to oranges. Indeed, it has been said that a meta-analysis is to an analysis what meta-physics is to physics. The interesting approach Stagg and colleagues took was to use mixed-treatment comparisons, also known as Bayesian network meta-analysis. This method allows for comparisons of different regimens when no trials directly compare them and provides an intelligent guess of the expected outcome if a trial was performed. Of course, such an estimate is not as high quality or authoritative as a direct trial. Given the multitude of studies on the treatment of LTBI, it is challenging for clinicians to compare the different treatment regimens. With the present study, Stagg and colleagues have brought considerable clarity to the issue.
Should the medical community retire isoniazid for 9 months as the standard therapy for LTBI? Based on current evidence the answer is "yes." It will be interesting to see if authorities like governmental and professional organizations switch their focus to short-course rifamycin-based regimens and author new clinical guidelines that reflect this shifting paradigm.