Population pharmacokinetics of lopinavir predict suboptimal therapeutic concentrations in treatment-experienced human immunodeficiency virus-infected children

Abstract

In adult protease inhibitor (PI)-experienced patients, a lopinavir (LPV) phenotypic inhibitory quotient (PIQ) of >15 has been associated with a higher likelihood of viral suppression. The aims of this study were to develop a population pharmacokinetic (PK) model of LPV in children and to estimate the probability of achieving a PIQ of >15. HIV-infected, PI-experienced children receiving LPV were intensively sampled for 12 h to measure plasma LPV. The data were fitted to candidate PK models (using MM-USCPACK software), and the final model was used to simulate 1,000 children to determine the probability of achieving an LPV PIQ of >15. In 50 patients (4 to 18 years old), the median LPV plasma 12-hour-postdose concentration was 5.9 mg/liter (range, 0.03 to 16.2 mg/liter) lower than that reported in adults. After a delay, LPV was absorbed linearly into a central compartment whose size was dependent on the weight and age of the patient. Elimination was dependent on weight. The regression line of observed versus predicted LPV had an R(2) of 0.99 and a slope of 1.0. Visual predictive checks against all available measured concentrations showed good predictive ability of the model. The probability of achieving an LPV PIQ of >15 was >90% for wild-type virus but <10% for even moderately resistant virus. The currently recommended dose of LPV/ritonavir appears to be adequate for children infected with wild-type virus but is unlikely to provide adequate inhibitory concentrations for even moderately resistant human immunodeficiency virus (HIV). PI-experienced HIV-infected children will likely benefit from longitudinal, repeated LPV measurement in plasma to ensure that drug exposure is most often near the maximal end of the observed safe range.

Figures

FIG. 1.

Individual observed LPV time-concentration curves for each of the 52 patient visits (gray lines). The median observed time-concentration curve for the entire population is superimposed (black line).

FIG. 2.

Linear regression of individual observed versus predicted LPV concentrations. Predictions using mean model parameter values (A) and using the means of the individual Bayesian posterior parameter distributions (B) are shown. The solid line is the fitted regression line. The dashed line is the unity line, which is fully superimposed on the regression line in panel B.

FIG. 3.

Model residual errors with respect to predicted LPV (A) and time (B).

FIG. 4.

Visual predictive check of concentration percentiles from model-simulated data (lines) superimposed on measured patient concentrations from the entire population (dots).

FIG. 5.

Distribution of lag times in the population.

FIG. 6.

Relationship between LPV concentrations and antiviral effect. (A) Percentages of children who will achieve a steady-state PIQ target of at least 15 when given the standard dose of 230 mg/m2 LPV/RTV twice daily for a given viral IC50. (B) Clinical activity of LPV plotted against increasingly resistant strains for the 25th, 50th, and 75th percentiles of LPV Ctrough concentrations in children given the standard dose.