Matters close to the heart: adaptation of tacrolimus models to inform therapeutic drug monitoring using the PRIOR approach

Introduction. External evaluation of published tacrolimus population pharmacokinetic models (n=17) highlighted that no available model accurately describes tacrolimus disposition in heart transplant (HTX) recipients, both with and without concomitant azole antifungal therapy.1 Concomitant azole antifungal therapy alters tacrolimus disposition substantially necessitating dose adjustments. The incorporation of azole antifungal use as a covariate may improve model-predicted tacrolimus exposure. The ability to optimise tacrolimus therapy in HTX recipients through model-informed precision dosing holds great promise to improve patient outcomes.

Aim. To develop and evaluate a population pharmacokinetic model for tacrolimus in HTX recipients, with and without concomitant azole antifungal therapy.

Methods. Data from HTX recipients in 2017 (1369 tacrolimus concentrations) and 2018 (1205 concentrations) administered the oral immediate-release formulation of tacrolimus (Prograf®) were obtained up to 391 days post-HTX. Due to data sparseness, the PRIOR approach was employed to support the estimation of pharmacokinetic parameters (absorption rate constant [Ka], lag time, central volume of distribution [V1], intercompartmental CL [Q], peripheral volume of distribution [V2], clearance [CL]) using values from two published models (Model 7 (M7)2 and Model 13 (M13)3)). These two models have previously displayed acceptable accuracy in predicting tacrolimus exposure in HTX recipients receiving concomitant azole antifungal therapy.1 Both models were implemented in NONMEM v7.4.3 and independently tested on data from HTX recipients in 2018 using the PRIOR NWPRI subroutine. The azole antifungal effect on tacrolimus CL and bioavailability were investigated. Internal evaluation of the updated models was assessed using prediction-corrected visual predictive check, pcVPC. The updated models were externally evaluated (NONMEM, FOCEI, MAXEVAL=0) using prediction-based metrics (bias and precision) and simulation-based metrics (pcVPC) using data from HTX recipients in 2017.

Results. Concomitant azole antifungal therapy influenced tacrolimus bioavailability for both updated M7 and M13. Pharmacokinetic parameter estimates for updated M7 were close to the prior values with the exception of CL (44 L/h vs. 21 L/h; original vs. updated model). Between-subject variability (BSV) in the V1 increased (144% vs. 535%). Internal and external evaluation (pcVPC) displayed that updated M7 inadequately described the data. In contrast, the pharmacokinetic parameter estimates for updated M13 were close to the prior values except for Ka (0.579 vs. 1.63). Internal evaluation showed that updated M13 adequately described the data. External evaluation of updated M13 using Bayesian forecasting approach displayed that the bias and precision with concomitant azole antifungal (1081 tacrolimus concentrations) were -3% (95%CI: -4.4−-1.7) and 13% (95%CI: 12.3−14.1) and, without concomitant azole antifungal (288 concentrations) were 0.1% (95%CI: -3.5−6.1) and 21% (95%CI: 18.0−24.0), respectively. The pcVPC adequately described the data.

Conclusion. Of the two models assessed, updated M13 adequately described tacrolimus disposition in HTX recipients. The incorporation of concomitant azole antifungal therapy as a covariate on bioavailability improved model performance. Prospective evaluation is required to assess the clinical utility of updated M13.

 

1Kirubakaran et al (2019) External evaluation of population pharmacokinetic models of tacrolimus in adult heart transplant recipients [Oral Presentation]. ASCEPT-PAGANZ Joint Scientific Meeting, Queenstown, New Zealand;

2Lu et al (2019) Br J Clin Pharmacol 85(8), 1692-1703;

3Sikma et al (2020) Eur J Drug Metab Pharmacokinet 45, 123-134.