Limited sampling strategy for the estimation of systemic exposure to the protease inhibitor nelfinavir.

Therapeutic drug monitoring (TDM) of antiretroviral drugs has been proposed as a means of optimizing response to highly active antiretroviral therapy (HAART) in HIV infection because suboptimal exposure to these agents may lead to the development of resistant viral strains and subsequent therapeutic failure. The area under the curve (AUC), though considered to make the best estimate of total drug exposure, requires repeated blood sampling. The authors investigated the predictability of individual nelfinavir (NFV) concentrations at different time points for the AUC and tried to find the best sampling time for the abbreviated AUC to predict NFV total body exposure. A total of 99 NFV AUC0-12h values were measured in 99 patients receiving a 1250-mg oral dose twice a day. Venous blood samples were collected at baseline (predose, 0) and 1, 2, 3, 4, 5, 6, 8, and 12 hours postdose. A stepwise forward-selection, multiple-regression technique was chosen to assess the relative importance of single and combination concentration time points to predict the AUC calculated from the entire pharmacokinetic profile. Data were split into a development set and a validation set. The development set contained 49 randomly selected HIV patients. Of these, 22 HIV patients were coinfected with HCV, 7 with and 15 without cirrhosis. One-point predictors provided the lowest prediction precision, but predictive performance improved after the first 2 hours postdose. Plasma concentrations at 0 and 4 hours after the oral dose were most predictive if 2 variables were used in the regression equation. The AUC could be estimated from data for these 2 samples by using the following equation: AUC0-12 = 3.0 + 2.7 (C0) + 6.4 (C4), r = 92. The predictive performance of 2-point predictors at 0 and 4 hours (C0 + C4) was validated by comparing their ability to predict the full AUC in a validation set representative of HIV/HCV patients (n = 28) and HIV/HCV patients, with (n = 8) and without (n = 14) cirrhosis. The results showed a mean bias ranging from +2.7% in HIV/HCV patients to -6.0% in HCV coinfection with cirrhosis. The authors conclude that this result is clinically significant. The limited sampling strategy (LSS) described could be used in clinical practice for the easy assessment of the total exposure to NFV in HIV/HCV patients, both with and without cirrhosis.