Comparative performance of oral midazolam clearance and plasma 4β‐hydroxycholesterol to explain interindividual variability in tacrolimus clearance

Aims We compared the CYP3A4 metrics weight‐corrected midazolam apparent oral clearance (MDZ Cl/F/W) and plasma 4β‐hydroxycholesterol/cholesterol (4β‐OHC/C) as they relate to tacrolimus (TAC) Cl/F/W in renal transplant recipients. Methods For a cohort of 147 patients, 8 h area under the curve (AUC) values for TAC and oral MDZ were calculated besides measurement of 4β‐OHC/C. A subgroup of 70 patients additionally underwent intravenous erythromycin breath test (EBT) and were administered the intravenous MDZ probe. All patients were genotyped for common polymorphisms in CYP3A4 , CYP3A5 and P450 oxidoreductase , among others. Results MDZ Cl/F/W, 4β‐OHC/C/W, EBT and TAC Cl/F/W were all moderately correlated ( r  = 0.262–0.505). Neither MDZ Cl/F/W nor 4β‐OHC/C/W explained variability in TAC Cl/F/W in CYP3A5 expressors ( n  = 29). For CYP3A5 non‐expressors ( n  = 118), factors explaining variability in TAC Cl/F/W in a MDZ‐based model were MDZ Cl/F/W ( R 2  = 0.201), haematocrit ( R 2  = 0.139), TAC formulation ( R 2  = 0.107) and age ( R 2  = 0.032; total R 2  = 0.479). In the 4β‐OHC/C/W‐based model, predictors were 4β‐OHC/C/W ( R 2  = 0.196), haematocrit ( R 2  = 0.059) and age ( R 2  = 0.057; total R 2  = 0.312). When genotype information was ignored, predictors of TAC Cl/F/W in the whole cohort were 4β‐OHC/C/W ( R 2  = 0.167), MDZ Cl/F/W ( R 2  = 0.045); Tac QD formulation ( R 2  = 0.036), and haematocrit ( R 2  = 0.032; total R 2  = 0.315). 4β‐OHC/C/W, but not MDZ Cl/F/W, was higher in CYP3A5 expressors because it was higher in CYP3A4*1b carriers, which were almost all CYP3A5 expressors. Conclusions A MDZ‐based model explained more variability in TAC clearance in CYP3A5 non‐expressors. However, 4β‐OHC/C/W was superior in a model in which no genotype information was available, likely because 4β‐OHC/C/W was influenced by the CYP3A4*1b polymorphism.