Parametric and nonparametric population pharmacokinetic analysis of fluconazole in critically ill patients and dosing simulations for Candida infections

Abstract

Large pharmacokinetic (PK) variability of fluconazole has been reported in critically ill patients, but the implications for fluconazole dosing remain unclear. The objectives of this study were to evaluate the population PK of fluconazole and identify appropriate dosage regimens by simulations. This was a retrospective analysis of fluconazole PK data from patients hospitalized in critical care and infectious disease departments. Both parametric and nonparametric population approaches were used. Various loading and maintenance fluconazole doses were evaluated by simulations, with computation of the probabilities of PK/pharmacodynamic (PD) target attainment (PTA) and cumulative fractions of response (CFR) based on international and local minimum inhibitory concentration (MIC) distributions of Candida sp. Data from 36 critically ill patients and 16 non-critically ill patients were available for model building (n = 202 concentrations). The final model adequately described the data, including the external data set (13 patients). After 24 h of therapy, 65% and 74% of patients had trough and area under the concentration-time curve values below the usual targets. Standard dosages were associated with low PTA for MIC >1 mg/L at 24 h. Higher loading doses administered two times daily improved PTA. CFR were >90% for C. albicans with standard dosages, while they were very low for C. glabrata, even with high dosages. Candida species and associated MIC distributions strongly influence fluconazole dosage requirements. Higher loading doses may be necessary for the achievement of PK/PD targets up to MIC breakpoints. The use of fluconazole for invasive C. glabrata infection should be discouraged because of poor PK/PD target attainment.

Keywords: antifungal; critical care; fluconazole; pharmacokinetics; therapeutic drug monitoring.

Fig 1

Population (left panel) and individual (right panel) predictions from the nonparametric model (Pmetrics) versus observed fluconazole concentrations in the learning (top) and validation (bottom) data sets. Units of observed and predicted concentrations are mg/L.

Fig 2

Prediction-corrected visual predictive checks from the final nonparametric model (Pmetrics) in the learning data set. Solid blue lines and shaded areas represent the 5th, 50th, and 95th percentiles of the observed data and the 90% confidence intervals of the model-based simulated observations (n = 1,000), respectively. Units are hours since the first fluconazole dose (x axis) and fluconazole concentration in mg/L (y axis).

Fig 3

Probability of target attainment for various loading and maintenance dosage of fluconazole in patients with normal renal function after 24 h and 7 days of therapy. PTA were calculated based on 1,000 simulated patients with CCR_IBW = 90 mL/min. The fluconazole MIC distributions of C. albicans and C. glabrata from EUCAST are shown as bars and expressed as proportions on the y axis. For ease of display, MIC values ≤0.125 and ≥16 mg/L are pooled in 0.125 and 16 mg/L category, respectively. Signs ×1 and ×2 indicate once-daily and twice-daily administration, respectively LD, loading dosage (day 1); MD, maintenance dosage.