Population pharmacokinetics of tacrolimus whole blood and peripheral blood mononuclear cell concentrations in stable kidney-transplanted patients

Abstract

Aim: Therapeutic drug monitoring of tacrolimus based on whole blood drug concentrations is routinely performed. The concentration of tacrolimus in peripheral blood mononuclear cells (PMBCs) is likely to better reflect drug exposure at the treatment target site. We aimed to describe the relationship between tacrolimus whole blood and PBMC concentrations, and the influence of patient characteristics on this relationship by developing a population pharmacokinetic model.

Methods: We prospectively enrolled 63 stable adult kidney-transplanted patients and collected dense (12-h, n = 18) or sparse (4-h, n = 45) pharmacokinetic profiles of tacrolimus. PBMCs were isolated from whole blood (Ficoll density gradient centrifugation), and drug concentrations in whole blood and PBMCs were analysed using liquid chromatography-mass spectrometry. Patient genotype (CYP3A4/5, ABCB1, NR1I2) was assessed with PCR. Population pharmacokinetic modelling and statistical evaluation was performed using NONMEM.

Results: Tacrolimus whole blood concentrations were well described using a two-compartment pharmacokinetic model with a lag-time and first-order absorption and elimination. Tacrolimus PBMC concentrations were best estimated from whole blood concentrations with the use of a scaling factor, the ratio of whole blood to PBMC concentrations (R_C:PBMC), which was the extent of tacrolimus distribution into PBMC. CYP3A5*1 non-expressors and NR1I2-25 385T allele expressors demonstrated higher R_C:PBMC ratios of 42.4% and 60.7%, respectively.

Conclusion: Tacrolimus PBMC concentration could not be accurately predicted from whole blood concentrations and covariates because of significant residual unexplained variability in the distribution of tacrolimus into PBMCs and may need to be measured directly if required for future studies.

Keywords: immunosuppressants; pharmacogenetics; population pharmacokinetics; therapeutic drug monitoring; transplantation.

FIGURE 1

Tacrolimus whole blood vs PBMC concentration in 63 kidney transplanted patients (R ² = 0.63, P < 0.001). Orange dots, significant erythrocyte contamination; green dots, no significant erythrocyte contamination.

FIGURE 2

Population pharmacokinetic model structure of the disposition of tacrolimus with a central, peripheral and intracellular compartment. CL, clearance; K _C‐PBMC, distribution rate constant from central compartment to PBMCs; MTT, mean transit time; R _C:PBMC, extent of tacrolimus distribution into PBMC; PBMC, peripheral blood mononuclear cells; Q, intercompartmental clearance; V _C, central volume of distribution; V _P, peripheral volume of distribution.

FIGURE 3

Prediction corrected visual predictive checks of (A) whole blood concentrations vs time, (B) PBMC concentrations vs time, (C) whole blood concentrations vs time using simulations based on the Andrews et al model and (D) PBMC concentrations vs time using simulations based on the Franken model. Grey dots, observed concentrations; dashed lines, observed 10th and 90th percentiles; solid lines, observed median concentrations; shaded area, 95% CI of the corresponding predicted percentiles; CI, confidence interval; PBMC, peripheral blood mononuclear cells; PK, pharmacokinetics.

FIGURE 4

Illustration of covariate relationships in relation to individual parameters of (A) relative bioavailability vs age, (B) central volume of distribution vs sex, (C) clearance vs CYP3A5*1/*3 genotype and (D) extent of distribution into PBMC vs CYP3A5*1/*3 and NR1I2‐25385C>T genotype. Black dots, individual predictions; green dashed line, typical individual predictions. The box represents the IQR of individual predictions with a solid black line indicating the median. The upper and lower whiskers represent 1.5 times IQR. IQR, interquartile range; PBMC, peripheral blood mononuclear cells.

FIGURE 5

Illustration of the covariate impact on pharmacokinetic profiles of tacrolimus in whole blood and PBMC for the covariate effect of (A) sex, (B) age, (C) NR1I2‐25385C>T genotype, (D) CYP3A5 expressors/non‐expressors and (e) the most extreme combinations of identified covariates. Based on 1000 simulations, where the line indicates the median and the shaded area indicates the 90% predictions intervals.