REF_iP-gp and REF_iBCRP Generation from Different Laboratories.

According to our literature analysis, human jejunum mRNA and protein expression was identified in 19 studies for P-gp and nine studies for BCRP [(Supplemental Materials; (Supplemental Table 1)]. Expression data for Caco-2 P-gp and BCRP from the same laboratory using the same protocol to generate an REF_iPgp or REF_iBCRP were found for five and four studies, respectively (Table 1). For P-gp, relative mRNA expression analysis (reverse-transcription polymerase chain reaction) enabled the generation of REF_iP-gp from two laboratories in three studies (Taipalensuu et al., 2001; Seithel et al., 2006; Hilgendorf et al., 2007), as the data from Seithel et al. (2006) and Hilgendorf et al. (2007) used Caco-2 monolayers cultivated in the same laboratory for 23 and 16 days, respectively. An REF_iP-gp from two independent laboratories that used Western blotting (Troutman and Thakker, 2003b; von Richter et al., 2009) is available, but not for LC-MS/MS quantification for either P-gp or BCRP, as the Caco-2 cell abundances reported by Oswald et al. (2013) were from plastic, not filter-grown cells (Dr. Stefan Oswald, personal communication). The REF_iP-gp based on mRNA expression ranged 7.1-fold, and for LC-MS/MS quantification, 2.6-fold. The REF_iBCRP of 90 (Taipalensuu et al., 2001) may result from low BCRP levels or variability in the housekeeping gene used (Seithel et al., 2006). The REF_iP-gp and REF_iBCRP generated from P-gp and BCRP quantification by LC-MS/MS from two different laboratories, UoM and Bertin Pharma (BPh), for the same samples are provided in Table 1 (Harwood et al., 2016). The REF_iP-gp (Troutman and Thakker, 2003b) from independent samples quantified by Western blotting was 5-fold higher than the REF_iP-gp generated by the UoM (UoM-REF_iP-gp), whereas the UoM-REF_iBCRP was approximately 2-fold higher than BPh (LC-MS/MS) and Altana AG (Western blot; Wesel, Germany) (von Richter et al., 2009).

Assessing the Sensitivity of REF_iP-gp in IVIVE-PBPK.

Both the UoM-REF_iP-gp of 0.4 (Harwood et al., 2016) and the REF_iP-gp of 2 (Troutman and Thakker, 2003b) led to digoxin C_max values within observed ranges after a single oral dose of 0.5 mg of digoxin (Fig. 1A), implying both REF_iP-gp reflect realistic contributions of P-gp in estimating observed C_max values when using the J_max and K_m for P-gp reported by Troutman and Thakker (2003a). Using the UoM-REF_iP-gp of 0.4 compared with the REF_iP-gp of 2 led to a modest 1.31- and 1.19-fold lower mean C_max and area under the curve, respectively, in 100 HV individuals (Fig. 1B). A previous study showed that the observed digoxin-rifampicin DDI, which was attributed to a 3.5-fold increase in intestinal P-gp expression (Greiner et al., 1999), could be recovered using an IVIVE-PBPK strategy, in which the REF_iP-gp of 2.0 was increased 3.5-fold to 7 after (Fig. 1C) induction (Neuhoff et al., 2013b). A 3.5-fold increase in the UoM-REF_iP-gp of 0.4 gave an REF_iP-gp of 1.4, leading to a simulated underprediction in the observed DDI (Fig. 1D). This indicates that the lower UoM-REF_iP-gp (that is not derived from the same Caco-2 system in which the apparent kinetic data were generated) is not sufficient to recover the contribution of P-gp induction by rifampicin to digoxin plasma concentration in HVs. The inability to recover the observed DDI when using the UoM-REF_iP-gp may result from lower P-gp expression and hence lower activity in the UoM Caco-2 systems. This can be due to laboratory differences in methods of expression quantification, Caco-2 cell cultivation, and the variability in jejunum expression. To recover the activity shortfall when using the UoM-REF_iP-gp, a 4.3-fold increase in J_max (1874 pmol/min/cm²) was required to recover the observed DDI (Fig. 1E) after using the Nelder-Mead minimization method and weighted least-squares algorithm in the simulators parameter estimation module (Jamei et al., 2014).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

(A) Sensitivity of C_max to REF_iP-gp for digoxin (single oral dose, 0.5 mg) in 100 HV individuals. The ranges of observed C_max values are given between the dashed lines, and the REF_iP-gp of 2.0 from Troutman and Thakker (2003b) (gray arrows) and the UoM-REF_iP-gp (Harwood et al., 2016) (dashed arrows) are shown. (B) The mean digoxin plasma concentration when using the REF_iP-gp of 2.0 (Troutman and Thakker, 2003b) and the UoM REF_iP-gp of 0.4 in 100 HVs. The observed and predicted plasma concentrations for digoxin (single oral dose, 1 mg) in 80 HV individuals after the dosing of rifampicin (600 mg, 11 doses, once daily) using an REF_iP-gp of 2 (Troutman and Thakker, 2003b) (C), REF_iP-gp of 0.4 (Harwood et al., 2016) (D), and REF_iP-gp of 0.4 (E) after optimizing the J_max of P-gp by parameter estimation. The thin gray lines represent mean values for 10 individual virtual trials of eight individuals, males aged 21–37 years; the thick black lines are the overall means of the virtual population (n = 80); and the dashed gray lines are the 95th and 5th percentiles of the confidence interval. Open circles mark the observed digoxin concentrations when coadministered with multiple doses of rifampicin (Greiner et al., 1999).

Assessing the Sensitivity of REF_iBCRP in IVIVE-PBPK.

The sensitivity of the region-specific fraction of dose absorbed and enterocyte concentrations to REF_iBCRP generated by BPh (1.11) and the UoM (2.22) (Table 1; Harwood et al., 2016) was assessed for the BCRP test compound TC. Figure 2 shows the free segmental enterocyte concentration for TC, used as the driving force for apical efflux transporters. As expected, the lower BPh-REF_iBCRP leads to higher TC enterocyte concentrations in proximal regions than the higher UoM-REF_iBCRP, whereas an increasing importance of intestinal BCRP UoM-(REF_iBCRP) results in higher TC absorption and higher enterocyte concentration in the distal intestine due to the efflux activity promoting TC retention in the gut lumen and transit to the colon, a region with 7.7-fold lower BCRP levels (Harwood et al., 2013). The higher REF_iBCRP has a limited impact on lowering C_max and area under the curve (1.22- and 1.03-fold, respectively), but increases t_max 1.44-fold to 2.8 hours and is in line with clinical observations, where an inhibition of intestinal BCRP leads to a decrease in t_max (Schneck et al., 2004). Alongside differences in BCRP expression, interindividual variability in system parameters, such as the small intestinal transit time (range 0.5–10 hours; Yu et al., 1996), also contributes to region-specific fraction of dose absorbed [Supplemental Fig. 1; (Supplemental Materials)]. Acidic BCRP substrates, such as rosuvastatin, are expected to possess higher enterocyte concentrations, as limited metabolism, low passive permeation, and apical uptake transporters operate (Li et al., 2012; Jamei et al., 2014). Therefore, increased BCRP expression alters t_max and regional absorption, while not limiting overall absorption and bioavailability. This is dissimilar to the cooperation of P-gp and CYP3A4 activities that facilitate a drug’s repeated exposure to intestinal CYP3A4, increasing overall gut metabolism and reducing bioavailability (Wacher et al., 1998). To our knowledge, the current study is the first highlighting this difference of the colocalized transporters P-gp and BCRP.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Simulated enterocyte concentration profiles for all intestinal segments of the model for the BCRP compound TC (10 mg oral, single dose, in solution) in 100 HVs using the BPh REF_iBCRP (1.11) or UoM-REF_iBCRP (2.22) with a mean small intestinal transit time of 3.34 hours, a passive apparent permeability of 115.2 × 10⁻⁶ cm/s, and an intrinsic clearance for BCRP of 17 µl/min/cm².

Combining IVIVE scalars and activity data generated from different laboratories for ATP-dependent transporters may not lead to successful IVIVE, whereas we postulate that laboratory-specific differences in REF may impact the mechanistic understanding of projected pharmacokinetic liabilities (efficacy/toxicity). This is due to in vitro activity, reproducibility of in vitro assays, culture conditions, and proteomic workflows. As discussed previously, direct translation of protein expression to activity may not always occur; therefore, accounting for deviations in this linear relationship via activity-abundance scalars will be required (Harwood et al., 2013). Ideally, scaling factors should be defined on a laboratory-specific basis against a common reference and combined with activity data from the same system. However, it is improbable within an industrial setting that groups will possess a bank of human intestinal tissues by which to obtain the in vivo abundance for in-house intestinal REF generation. It is therefore advocated that commercially available pooled human intestinal microsomes (constituting ≥20 intestines) are used to generate an REF using the same proteomic methods as those used for quantifying in vitro system abundances used for determining activity. Alternatively, a link between human liver microsomes, intestinal microsomes, and Caco-2 cells can be approached.