CLINICAL IMMUNOSUPPRESSION - KIDNEY EARLY

K.A. Barraclough¹, N.M. Isbel¹, K.J. Lee², D.W. Johnson³, D. Leary⁴, P.J. Taylor⁵, S.B. Campbell¹, C.E. Staatz^6
1Nephrology, Princess Alexandra Hospital, Brisbane/AUSTRALIA, ²Pharmacy, University of Queensland, Brisbane/QLD/AUSTRALIA, ³Department Of Nephrology, University of Queensland at Princess Alexandra Hospital, Brisbane/AUSTRALIA, ⁴Nephrology, Princess Alexandra Hospital, Brisbane/QLD/AUSTRALIA, ⁵Medicine, Princess Alexandra Hospital, Brisbane/AUSTRALIA, ⁶School Of Pharmacy, University of Queensland, Brisbane/AUSTRALIA

Body: Introduction: Multiple regression derived limited sampling strategies (LSSs) for estimation of tacrolimus area under the concentration-time curve from 0 to 12 hours post dose (AUC0-12) may bepreferable to pre-dose (C0) concentrations for tacrolimus therapeutic drug monitoring (TDM). The aim of this study was to test the predictive performance of all published multiple regression derivedLSSs for tacrolimus in an independent population of adult kidney transplant recipients. Methods: Six articles presenting 42 multiple regression-derived LSSs for tacrolimus (Prograf®,Janssen-Cliag), were identified from the literature. Twenty tacrolimus AUC0-12 profiles, comprising 13 concentration-time measurements, were collected from 20 adult kidney transplant recipients.Study participants were divided into early (3-5 days post-transplant surgery; study group 1) and late (> 3 months post-transplant surgery; study group 2) cohorts. Full tacrolimus AUC0-12 (AUCf)was calculated from all measured concentration-time points using non-compartmental analysis (trapezoidal rule). Predicted tacrolimus AUC0-12 (AUCp) was calculated by applying relevant concentrationmeasurements within each of the multiple regression equations identified. A Pearson correlation coefficient test was applied to access the correlation between 1) C0 and AUCf and 2) AUCp and AUCf.Four error indices [median prediction error (MPE), median percentage prediction error (MPPE), root median squared prediction error (RMSE) and median absolute percentage prediction error (MAPE)] wereused to evaluate bias and precision. Results: Median (IQR) dose-adjusted tacrolimus AUCf was higher in those > 3 months post-transplant compared with those in their first post-transplant week[41.2 (33.9, 56.6) mcg·h/L/mg tacrolimus versus 20.2 (10.7, 26.5) mcg·h/L/mg tacrolimus; p=0.002]. The r2 value for the correlation of C0 with AUCf was 0.53. The correlation was lowerin the late compared with the early post-transplant group [r2 0.21 versus 0.64]. Based on the 42 multiple regression equations, correlation (r2) between AUCp and AUCf ranged from 0.053 to 0.99. Allequations showed higher correlation with AUCf than did C0. MPPE varied from -0.7% to 33.5%, and MAPE varied from 2.0% to 33.5%. MPPE and MAPE were < 15% for 29 of the 42 equations (62%). Anequation that used concentration measurements collected pre-dose and at 1, 2 and 4-hours post-dose (AUC0-12= -5.385 + 3.337C0 + 0.96C1 + 1.402C2 + 6.01C4) was superior to all other equations withregard to practicality and performance (sampling confined to 4 hours post-dose, r2 0.95, MPPE -1.1% and MAPE 2.8%). This equation remained superior regardless of duration post-transplant (study group1: r2 0.92, MPPE -0.8% and MAPE 2.1%; study group 2: r2 0.91, MPPE -1.9% and MAPE 3.1%). The two equations that incorporated only C0 concentration measurements displayed the lowest correlations,greatest bias and poorest precision of all of the equations. Conclusions: Multiple regression-derived LSSs have superior ability to predict tacrolimus AUCf as compared to C0 monitoring. While regularuse of LSSs may not be feasible in the early post-transplant period when frequent TDM is required, they may be of particular use later post-transplant when the relationship between C0 and AUCf isespecially poor.

Disclosure: All authors have declared no conflicts of interest.