OBJECTIVE: To determine the frequency and magnitude of below normal ap
parent peak serum concentrations for antituberculosis drugs in patient
s with AIDS and CD4 cell counts less than 200 cells/mm(3). We also exp
lored the data for potential relationships between response variables
and patient characteristics. DESIGN: Prospective study of consecutive
patients seen in tuberculosis clinics. SETTING: Five urban tuberculosi
s clinics in four major metropolitan areas. PARTICIPANTS: Twenty-six p
atients diagnosed with HIV infection and receiving treatment for activ
e tuberculosis were eligible. MAIN OUTCOME MEASURES: After 2 weeks or
more of therapy, blood was collected 2 hours after observed doses of t
he antituberculosis drugs. Serum samples were frozen, shipped to Natio
nal Jewish Center in Denver, and analyzed by HPLC or GC. Serum concent
rations were compared with the proposed normal ranges. Data were analy
zed to determine correlations between antituberculosis drug serum conc
entrations and patient characteristics. RESULTS: LOW 2-hour serum conc
entrations were common for antituberculosis drugs, particularly rifamp
in and ethambutol. Absorption of isoniazid was generally high. Potenti
al drug-drug interactions were found between rifampin and fluconazole
(fluconazole appears to increase rifampin concentrations) and between
pyrazinamide and zidovudine (zidovudine may lower pyrazinamide concent
rations). Patients receiving pyrazinamide had lower rifampin concentra
tions than those not receiving pyrazinamide. CONCLUSIONS: Low antitube
rculosis drug serum concentrations occur frequently during the treatme
nt of tuberculosis in patients with AIDS. Additional research is requi
red for patients with drug-resistant tuberculosis, and to clarify the
nature of the potential drug-drug interactions.