Tuberculosis Drug Serum Levels

Sir—The fine paper by Narita and colleagues [1] does not show a direct correlation between the serum concentrations of the tuberculosis drugs and the recurrence of tuberculosis disease. It is possible that the use of different break-point values (e.g., one-half of the target ranges) might have yielded different results. Ongoing research sponsored by the Centers for Disease Control and Prevention (Atlanta) suggests that, in certain situations, low serum concentrations of the tuberculosis drugs are associated with worse treatment outcomes. Some of these data were presented at the annual meeting of the American Thoracic Society in May 2001 [2, 3]. The available data suggest that failure of tuberculosis treatment or recurrence of disease results from a complex and poorly understood combination of factors that likely varies from patient to patient. The use of antibiotic serum concentrations rarely has perfect predictive value to guide the choice of drug dosage [4-10]. I would like to suggest an approach to the use of antibiotic serum concentrations that may be useful. First, it is clear that some drug concentration is required to achieve a therapeutic effect, and target ranges should be designed with a margin of safety with respect to their efficacy. Thus many “below-range” patients should be within this margin of safety. Second, the concentrations required for patients who are receiving multiple antibiotics may be lower than the target ranges for the individual drugs [11]. Third, for certain patients, achieving critical concentrations is the difference between success and failure [12, 13]. However, it is very difficult to identify these patients a priori, and it is highly unlikely that the same target concentrations will be successful in all situations. Typically, the pharmacodynamic relationship between serum drug concentrations (X) and therapeutic efficacy (Y) creates an S-shaped curve that is described mathematically by Hill equations [14]. Fourth, once the decision is made to use a given antibiotic, a goal should be set for the desired serum concentrations, and the goal should be achieved with the greatest precision possible [15]. That is where carefully timed serum concentrations are particularly helpful. Fifth, there are separate probability curves for serum drug concentrations (X) and selected drug toxicities (Y). These relationships also must be considered when setting serum concentration goals. Furthermore, if the disease state is potentially fatal, some toxicity may need to be accepted to effect a cure. Sixth, the decision to determine serum concentrations during the course of treatment is the decision to change the dose if necessary. Some patients will require doses greater than the “usual” or “recommended” doses. I would suggest that treatment guidelines avoid the term “maximum” [16]. The “maximum” dose for any patient is the dose that produces the desired therapeutic effect with an acceptable degree of toxicity. In addition, our concept of what constitutes “usual” doses needs to be adjusted in light of the possibility of drug-drug interactions, the degree of which can be unpredictable in an individual patient [17]. One is unlikely to find a perfect relationship between serum concentrations (1 variable) and treatment outcomes, which depend on multiple variables. However, we do know that treatment fails in selected patients because of inadequate serum concentrations. Therefore, if patients with tuberculosis fail to respond appropriately to directly observed treatment, monitoring of serum concentrations and repeated susceptibility testing should be performed. Critically ill patients with tuberculosis should receive as many tuberculosis drugs as possible intravenously (i.e., isoniazid, rifampin, streptomycin or amikacin, and levofloxacin). For the tuberculosis drugs that cannot be given intravenously, serum concentrations should be checked as soon as possible, and the dosage adjusted as needed. Finally, for the treatment of multiple drug-resistant tuberculosis, the achievable serum concentrations of the “second-line” tuberculosis drugs barely exceed the amounts needed to inhibit mycobacteria [10]. Therefore, I would advocate checking the serum concentrations and adjusting the drug dosages before embarking on 30 months of potentially toxic treatment. In conjunction with sound clinical judgment, therapeutic drug monitoring allows the clinician considerable flexibility in managing even the most difficult cases of infectious disease, including tuberculosis and multiple drug-resistant tuberculosis.