T.C. Zwart, A.G.J. Engbers, R.E. Dam, P.J.M. van der Boog, J.W. de Fijter, H.J. Guchelaar, A.P.J. de Vries, D.J.A.R. Moes
Thursday 5 march 2020
13:50 - 14:00h at Theaterzaal
Parallel session: Parallel sessie XII – Klinische en Basale abstracts
Background: In certain clinical situations, creatinine clearance (CrCl)-based glomerular filtration rate (GFR) estimation lacks accuracy for renal function evaluation. Measured GFR by determination of the plasma clearance of the exogenous marker iohexol may then provide an alternative. The need for 5-10 samples up to 8h postdose has however hampered its widespread use in routine clinical care. We developed a population pharmacokinetic (PK) and limited sampling model (LSM) to enable pragmatic iohexol-based GFR estimation in the renal transplant setting.
Methods: Blood samples (n = 328) drawn at 5 min to 4 h after iohexol administration were available from 30 renal transplant donors and 19 recipients. A population pharmacometric model was constructed to describe iohexol PK. Model evaluation was performed using diagnostic plots, visual predictive checks and bootstrap analysis. The final model was applied to develop LSMs based on 1-4 samples drawn within 3h postdose to find a sampling scheme which ensured accurate GFR estimation and clinical feasibility. GFR estimations of each LSM (GFRlsm) were compared to those of the full model (GFRfull) to evaluate LSM predictive performance. Predictive performance was assessed using the Pearson R2, mean percentage prediction error (MPPE), root mean squared prediction error (RMSE) and the percentage of GFRlsm within ±5% of the GFRfull (5%-discordance).
Results: Iohexol PK were best described by a 2-compartmental first order elimination model. Clearance (CL), intercompartmental clearance (Q) and distribution volumes of the central (Vc) and peripheral (Vp) compartments were 4.89 L/h (6% residual standard error; RSE), 7.26 L/h (25% RSE), 9.20 L (6% RSE) and 5.48 L (14% RSE). Interpatient variability of CL, Q, Vc and Vp was 34.4% (14% RSE), 86.2% (18% RSE), 35.2% (12% RSE) and 41.7% (44% RSE). Internal model evaluation indicated good performance. LSMs using three or more samples including an early and a late sample yielded the most accurate GFR estimations. The LSM with sampling at 5min, 1h, 2h and 3h showed the optimal combination of predictive performance (R2: 0.997; MPPE: 0.36%; RMSE: 0.39%; 5%-discordance: 95.92%) and clinical feasibility.
Conclusions: Our population PK model and LSM provide a promising approach to enable pragmatic iohexol plasma clearance-based GFR estimation in the renal transplant setting. External validation of the model is however warranted before implementing this technique in routine clinical care.