Article Figures & Data
Figures
- Fig. 1.
Workflow of the development and application of the PBPK kidney model for digoxin.
- Fig. 2.
Mean simulated digoxin plasma concentration time profiles (intravenous administration of 1 mg digoxin) for intermediate PBPK models used during development of the mechanistic kidney model. CLR defined by a single input value (136.1 ml/ min) based on the literature analysis (orange line) and CLR simulated using the mechanistic kidney model, accounting for only glomerular filtration (purple line), glomerular filtration and reabsorption (turquoise line), or glomerular filtration, reabsorption, and active secretion (blue line).
- Fig. 3.
Simulated digoxin CLR (A) and Cmax, PT-1 (B) at different input values for the kidney transporter kinetic parameters. Values of OATP4C1 CLint,T and P-gp REF were varied using the automated sensitivity analysis tool in the SimCYP simulator in a population representative following the clinical trial design reported previously (Greiner et al., 1999). Insets show the graphs presented on logarithmic scales.
- Fig. 4.
Estimation of OATP4C1 CLint,T parameter using a sensitivity analysis approach by simulating digoxin CLR in population representatives with different serum creatinine values. Colored meshes and gray horizontal plane indicate the simulated CLR and the overall weighted (by subject number) mean CLR obtained from the literature analysis (136.1 ml/min; n = 214 healthy subjects), respectively. Values of OATP4C1 CLint,T and serum creatinine parameters were varied using the automated sensitivity analysis tool in the SimCYP simulator. Optimal OATP4C1 value was taken at the intersection (yellow star) of the simulated digoxin CLR with the observed CLR at a serum creatinine value of 80 µmol/l (which corresponds to simulated GFR ∼120 ml/min), as indicated by the blue arrows. Sensitivity analysis was performed twice using clinical trial designs reported previously (Kramer et al., 1979; Rengelshausen et al., 2003).
- Fig. 5.
Comparison of simulated and observed digoxin plasma concentration and urinary excretion rate profiles using SimCYP with MechKiM after 0.5 mg by 1 hour infusion (Ochs et al., 1978)(A), 1 mg by 1 hour infusion (Ochs et al., 1978)(B), 1.5 mg by 1 hour infusion (Ochs et al., 1978)(C), 0.75 mg by bolus injection (Johnson and Bye, 1975)(D), and urinary excretion profiles only after 0.4 mg by bolus injection (Lindenbaum et al., 1981)(E) and 0.4 mg by 1 hour infusion (Lindenbaum et al., 1981)(F). Mean (purple solid lines) and 5th and 95th percentiles (dashed purple line) of simulated plasma concentrations are overlaid with mean observed data (purple circles), whereas mean simulated urinary excretion rates (gray solid lines) are overlaid with mean observed data (open circle).
- Fig. 6.
Simulated CLR in comparison with GFR and OATP4C1 abundance in virtual populations. (A) Simulated CLR and GFR data (purple open circle) in healthy and moderate and severe renal impairment virtual subjects, in comparison with reported clinical data (CLCR data on horizontal axis) (black X) (Bloom et al., 1966; Okada et al., 1978). Solid black line represents linear line of best fit using total least squares regression (which recognizes experimental error in both variables) for the observed clinical data. (B) Simulated CLR and GFR in healthy (yellow solid squares) and moderate (green open circles) and severe (turquoise solid diamond) renal impairment virtual subjects. Solid lines represent linear lines of best fit using ordinary least squares regression for data from each simulation, with relevant equations and R2 shown in boxes.
- Fig. 7.
Impact of reduced renal secretion on simulated digoxin AUC ratio (A), CLR (B) and Cmax,PT-1 ratio (C) in renal impairment populations. Renal secretion was reduced either by changing the kidney OATP4C1 or P-gp relative abundance parameters or by reducing the PTCPGK parameter by a proportional amount. Lines represent changes in PTCPGK in moderate renal impairment (light green solid line) and severe renal impairment (green dashed line), OATP4C1 abundance in moderate renal impairment (light blue solid line) and severe renal impairment (blue dashed line), and P-gp abundance in moderate renal impairment (light purple solid line) and severe renal impairment (purple dashed line). Each scenario was simulated in 100 virtual subjects. Solid horizontal black line (ratio = 1) represents the healthy volunteer population; estimated CLR ratios for the average moderate (GFR = 46.5 ml/min/1.73 m2; orange solid line) and severe (GFR = 23.5 ml/ min/1.73 m2; orange dashed line) renal impairment were calculated based on the correlation of GFR and CLR in the observed data (Bloom et al., 1966; Okada et al., 1978) (A). Relative change of PTCPGK or transporter abundance of 1 indicates that the default moderate or severe renal impairment population in the SimCYP simulator was used
- Fig. 8.
Simulation of digoxin CLR in population representative mode with changes in different systems parameters performed to represent changes in the case of renal impairment. Glomerular filtration rate (range 20–140 ml/min/1.73 m2) was changed by altering the serum creatinine parameter (74.5–695.6 µmol/l); OATP4C1 abundance and PTCPGK parameters were altered by a factor proportional to the relative change in GFR from the population representative of the default "healthy volunteers” population (GFR = 136.4 ml/min/1.73 m2; serum creatinine = 76.5 µmol/l). Lines represent simulations performed with changes in GFR alone (orange dashed line), both GFR and OATP4C1 abundance (blue dashed line), or both GFR and PTCPGK (green dashed lines). Reported clinical data (black X) are overlaid (Bloom et al., 1966; Okada et al., 1978).
Tables
- TABLE 1
Parameters used to simulate digoxin CLR
Reduction in filtration and secretion was performed to represent changes in renal impairment. Simulated population representative of the “healthy volunteers” population had an age, weight, and body surface area of 20 years, 81 kg, and 1.98 m2, respectively. Serum creatinine (input parameter of model) was calculated for each scenario using the Cockcroft-Gault equation (Cockcroft and Gault, 1976), based on the target GFR and the age, weight, and body surface area of the population representative.
GFR Serum Creatinine Concentration OATP4C1 Abundance PTCPGK ml/min/1.73 m2 µmol/l million PTC/g kidney 136.4a 76.5 1 60 140 74.5 1.03 61.6 120 86.9 0.88 52.8 100 104.3 0.73 44.0 80 130.4 0.59 35.2 60 173.9 0.44 26.4 40 260.8 0.29 17.6 20 521.7 0.15 8.8 15 695.6 0.11 6.6 ↵a Relative change in GFR for each scenario was calculated using the value of 136.4 ml/ min/m2 as baseline and applied to the OATP4C1 abundance or PTCPGK parameter.
- TABLE 2
Values of the OATP4C1 CLint,T parameter estimated by various methods and subsequent simulated digoxin CLR in healthy volunteers using the intravenous trial design from Greiner et al. (1999)
Source of Data OATP4C1 CLint,T Value Simulated Digoxin CLR µl/min/million PTC ml/min In vitro–in vivo extrapolation MDCK-OATP4C1 (Mikkaichi et al., 2004) 0.23 90.6 CHO-OATP4C1 (Chu et al., 2007) 270 804.3 Parameter estimation: Fitting to plasma concentration-time profile 1 mg, i.v. bolus (n = 12 subjects) (Kramer et al., 1979) 8.20 0.01 mg/kg, i.v. 4 min infusion (n = 12 subjects) (Rengelshausen et al., 2003) 1.10 0.5 mg, i.v. 5 min infusion (n = 12 subjects) (Ding et al., 2004) 0 0.75 mg, i.v. bolus (n = 8 subjects) (Koup et al., 1975) 0.09 0.75 mg, i.v. 1 h infusion (n = 8 subjects) (Koup et al., 1975) 5.56 0.75 mg, i.v. 3 min infusion (n = 8 subjects) (Johnson and Bye, 1975) 3.30 0.5 mg, i.v. 1 h infusion (n = 9 subjects) (Ochs et al., 1978) 0 1 mg, i.v. 1 h infusion (n = 9 subjects) (Ochs et al., 1978) 0 1.5 mg, i.v. 1 h infusion (n = 9 subjects) (Ochs et al., 1978) 0 Overall weighted mean ± standard deviation 1.85 ± 2.87 108.8 Parameter estimation: Sensitivity analysis Overall weighted mean CLRa (Population representative using trial design 1 mg, i.v. bolus (Kramer et al., 1979)) 4.14 133.4 Overall weighted mean CLRa (Population representative using trial design 0.01 mg/ kg, i.v. 4 min infusion (Rengelshausen et al., 2003)) 4.14 133.4 PTC, proximal tubule cells; MDCK, Madin-Darby canine kidney.
↵a CLR data presented in Supplemental Table S4.
Description (units) Value Comment fu,kidney 0.51 Predicted (Rodgers et al., 2005; Rodgers and Rowland, 2006) fu,urine 1 OATP4C1 CLint,T (µl/min/million PTC) 4.14 Estimated using sensitivity analysis Allocated to OAT1/ SLC22A6 transporter in MechKiM as OATP4C1 not defined in model OATP4C1 RAF/ REF 1 P-gp Km (µM) 177 Same as liver/gut (Neuhoff et al., 2013b) P-gp Vmax (pmol/min/million PTC) 434 Same as liver/gut (Neuhoff et al., 2013b) P-gp REF 1.51 Calculated from mRNA expression data (Hilgendorf et al., 2007) CLPD (µl/min/million PTC) 0.01 Estimated using sensitivity analysis and comparing simulated Freab with that predicted using static tubular reabsorption model (Scotcher et al., 2016c) and published Caco-2 data (Neuhoff et al., 2003; Zhang and Morris, 2003; Djuv and Nilsen, 2008; Fossati et al., 2008). Same value for apical and basolateral membranes in all segments of nephron fu,kidney, fraction unbound in kidney; fu,urine, fraction unbound in urine; Freab, fraction reabsorbed; CLint,T, transporter intrinsic clearance; CLPD, permeability diffusion clearance; Km, Michaelis constant; MechKiM, mechanistic kidney model in SimCYP simulator; PTC, proximal tubule cells; RAF, relative activity factor; REF, relative activity factor; Vmax, maximal velocity.
- TABLE 4
Simulated digoxin CLR and AUC0-∞ parameters in different virtual populations
Each value represents the mean of 100 simulated individuals taken from the virtual population provided with the SimCYP simulator after a single 0.75 mg i.v. dose.
Population Simulated CLR % of healthy CLR Simulated AUC0-∞ % of healthy AUC0-∞ ml/min µg.min/ml Healthy 131.9 100 3.36 100 Elderly 90.7 69 5.12 152 Moderate renal impairment 65.9 50 5.68 169 Severe renal impairment 43.8 33 7.00 208
Data Supplement
Files in this Data Supplement:
- Supplemental Data - 5 supplemental tables, 4 figures.
In this issue
PBPK Kidney Model of Digoxin and Renal Impairment
