. 2016 Oct 30;94:59–71. doi: 10.1016/j.ejps.2016.03.018

Table 4.

Assessment of the predictive performance of various CL_R prediction methods using gmfe and % predicted within 3-fold of observed CL_R.

	gmfe (% predicted within 3-fold of observed)
	Filtration only^a	No correction for microvilli^b	Proximal tubule only^c	Reabsorption model^d	P_app–F_reab′ calibration^e
All drugs (n = 45)	3.73 (58%)	5.35 (27%)	2.17 (76%)	1.96 (87%)	1.65 (91%)
Low F_reab (n = 17)	1.17 (100%)	5.02 (35%)	1.59 (94%)	1.97 (88%)	1.34 (94%)
Medium F_reab (n = 12)	2.56 (75%)	8.52 (17%)	1.44 (92%)	1.90 (92%)	1.73 (92%)
High F_reab (n = 16)	16.86 (0%)	4.03 (25%)	4.11 (44%)	2.01 (81%)	1.98 (88%)

CL_R,filt calculated using Eq. (7) in main text.

No correction was made for surface area attributable to the presence/absence of microvilli when calculating CL_R,int,reab,i.

CL_R predicted using a model with only one tubular compartment representing proximal tubule (main contributor to reabsorption predicted by the model).

CL_R predicted using tubular reabsorption model, as per Eq. (10), (12).

CL_R predicted using the tubular reabsorption model after calibration of P_app data using Eq. (14), data are for all drugs including reference subset.