TABLE 2

Accuracy and bias in the prediction of clinical CYP3A4 induction using various static models

The number of false negatives for each prediction method is listed. In parenthesis is the total number of predictions in the data set. For the R₃ model, an R value <0.9 is considered to indicate induction risk. For the mechanistic static, induction potential is considered possible if the AUCR value is <0.8.

Lot	Nominal EC₅₀^a	Measured EC₅₀^b	Nominal EC_50,u^c	Measured EC_50,u^d
RIS model
RMSE
NON	9.09	6.25	6.25	6.25
Hu1624	7.40	11.40	6.25	6.25
CDP	6.66	10.39	6.25	6.25
GMFE
NON	1.58	1.47	1.52	1.52
Hu1624	1.53	1.74	1.52	1.52
CDP	1.49	1.68	1.52	1.52
False negative (N)
NON	0 (7)	0 (7)	0 (6)	0 (6)
Hu1624	0 (7)	0 (7)	0 (6)	0 (6)
CDP	0 (7)	0 (7)	0 (6)	0 (6)
Basic static (R₃)
RMSE
NON	20.84	18.35	16.05	15.88
Hu1624	19.67	16.53	16.22	15.43
CDP	26.02	24.58	23.76	23.38
GMFE
NON	2.16	2.13	2.29	2.29
Hu1624	1.99	1.92	2.06	2.04
CDP	2.34	2.30	2.48	2.46
False negative (N)
NON	2 (7)	1 (7)	0 (6)	0 (6)
Hu1624	2 (7)	1 (7)	1 (6)	0 (6)
CDP	2 (7)	1 (7)	1 (6)	1 (6)
Mechanistic static with inhibition
RMSE
NON	19.64	15.77	11.99	11.50
Hu1624	19.29	15.94	14.97	13.20
CDP	24.02	22.00	20.88	19.89
GMFE
NON	2.93	2.98	3.42	3.46
Hu1624	2.36	2.24	2.53	2.67
CDP	3.19	3.27	3.79	3.79
False negative (N)
NON	2 (7)	2 (7)	1 (6)	0 (6)
Hu1624	2 (7)	1 (7)	1 (6)	1 (6)
CDP	2 (7)	2 (7)	1 (6)	1 (6)

↵^a Nominal EC₅₀ is the EC₅₀ value determined based on the nominal concentration of dosing solution (total C₀).
↵^b Measured EC₅₀ is the EC₅₀ value determined based on the measured concentration in induction medium (total C_ave): measured EC₅₀ = nominal EC₅₀ × C_ave/C₀.
↵^c Nominal unbound EC₅₀ (EC_50,u) is the EC₅₀ value determined based on the unbound concentration of dosing solution (C_0,u): unbound EC₅₀ = nominal EC₅₀ × f_u (medium).
↵^d Measured EC_50,u is the EC₅₀ value determined based on the measured unbound concentration in induction medium (C_ave,u): measured EC_50,u = nominal EC₅₀ × f_u (medium) × C_ave/C₀.