(T1130-02-07) Novel Method of Quantitatively Predicting the Clearance of Uridine Diphosphate-Glucuronosyltransferase Substrates Using Human Liver Chimeric Mice

Purpose: Clearance is an essential parameter for the construction of physiologically-based pharmacokinetic models. Uridine diphosphate-glucuronosyltransferases (UGTs) is a superfamily of enzymes that metabolize a variety of xenobiotics, including medicines, bile acids, bilirubin, thyroid hormones, steroid hormones, fat-soluble vitamins, and neurotransmitters. Notably, the UGT1 and UGT2 families, which are expressed in the liver, intestine, and kidney, play a critical role in drug metabolism. Furthermore, UGTs make the most significant contribution to drug metabolism outside of cytochrome P450 enzymes¹. Thus, accurately predicting the clearance of UGT substrates is crucial for drug discovery and development. However, current in vitro systems and animal scale-up approaches often fail to predict clearance accurately. A liver-specific humanized mouse model transplanted with human hepatocytes represents a promising tool for predicting human drug disposition mediated by metabolic enzymes, transporters, as well as for assessing toxicity. Consequently, we hypothesized that these liver-specific humanized mice can more accurately predict clearance for UGT substrates than previous models. The present study aimed to evaluate whether hu-PXB mice, a type of liver-specific humanized mouse, can improve the predictability of UGT-mediated clearance in humans.

Methods: To estimate human hepatic intrinsic clearance (CL_h,int) in vitro, UGT substrate drugs (acetaminophen, entacapone, ketoprofen, lorazepam, oxazepam, posaconazole, and zidovudine) were incubated with human hepatocytes. CL_h,int was calculated based on the rate of substrate disappearance and free fraction in hepatocytes. In vivo human CL_h,int values were calculated based on published data. To evaluate human CL_h,int predictability, the seven substrate drugs were administered intravenously to hu-PXB and SCID mice. Blood samples were collected from the jugular vein using a heparinized syringe at 0.033, 0.167, 0.5, 1, 2, 4, 7, 24 and 48 (SCID mice only) hours after substrate administration. Urine and faecal samples were also collected for 24 and 48 (SCID mice only) hours after intravenous administration. The in vitro and in vivo samples were analyzed by LC–MS/MS. We calculated the CL_h,int in the mice. We compared the CL_h,int in humans with that in human hepatocytes, hu-PXB, and SCID mice (Fig. 1). To evaluate UGT-mediated drug-drug interaction, lorazepam as a verifiable compound was administered intravenously to hu-PXB mice with probenecid as a UGTs inhibitor.

Results: The plasma concentration–time profiles of the seven substrate drugs and their glucuronide metabolites in hu-PXB and SCID mice are shown in Fig. 2A. The pharmacokinetic parameters obtained in the present study are summarized in Fig 2B. The CL_total and CL_h of ketoprofen, lorazepam, oxazepam, posaconazole, and zidovudine in hu-PXB mice were larger than those of SCID mice. On the other hand, the CL_total and CL_h of acetaminophen in hu-PXB mice were smaller than those of SCID mice. The posaconazole faecal excretion were much larger, exceeding 60% of injected doses in hu-PXB and SCID mice. The urine and faecal excretion of lorazepam and oxazepam in SCID mice was much larger than those of hu-PXB mice. We also measured the metabolite concentrations of the seven substrates in urine and faeces of hu-PXB and SCID mice. Acetaminophen D-glucuronide, entacapone 3-β-D-glucuronide, ketoprofen Acyl-β-D-glucuronide, lorazepam glucuronide, oxazepam glucuronide, and zidovudine O-β-D-glucuronide were mainly excreted into urine. The values of lorazepam, oxazepam, posaconazole, and zidovudine in hu-PXB mice were much higher than those of SCID mice. In contrast, the value of acetaminophen in hu-PXB mice was lower than SCID mice. The correlation of human CL_h,int between in vitro and in vivo was relatively good (Fig. 3A). However, the in vitro to in vivo extrapolation in the present study predicted the CL_h,int of the UGT substrate drugs within twofold of the clinical values for 14%. We investigated the relationship in calculated CL_h,int among humans and both mice in Fig. 3B and 3C. The figures show a much better correlation in CL_h,int between hu-PXB mice and humans than that between SCID mice and humans. Hu-PXB mice predicted the CL_h,int of UGT substrate drugs within twofold of the clinical values for 100%. In contrast, SCID mice predicted it within twofold of the clinical values for 43%. Comparing the prediction accuracy of UGT substrate CL_h,int with that of other reports, Hu-PXB mice were found to predict CL_h,int of UGT substrate drugs the most accurately^2-3. Moreover, %change in CL_total of lorazepam inhibited by probenecid in Hu-PXB mice corresponded to that in humans (Fig. 3D).

Conclusion: Our findings demonstrate that liver specific humanized mice can accurately predict the clearance of UGT substrates and may be capable of predicting UGT-mediated drug-drug interaction. This in vivo approach surpasses previously reported methods in accurately predicting the disposition of drugs primarily metabolized by UGTs. This new humanized animal platform will address major limitations in human pharmacokinetics prediction.

References: 1) Saravanakumar et al., 2019
2) Deguchi et al., 2011
3) Docci et al., 2020


Fig. 1. Outline of the human prediction for UGT substrate clearance using human liver chimeric mice


Fig. 2. (A) Plasma concentration–time profiles of the seven substrate drugs and their glucuronide metabolites in hu-PXB and SCID mice. Mice were intravenously administered acetaminophen (20 mg⋅kg−1), entacapone [4 mg⋅kg−1 (Hu-PXB) or 2 mg⋅kg−1 (SCID)], ketoprofen (10 mg⋅kg−1), lorazepam (0.4 mg⋅kg−1), oxazepam (3mg⋅kg−1), posaconazole (10 mg⋅kg−1), and zidovudine (10 mg⋅kg−1). (B) Pharmacokinetic parameters of the seven substrate drugs in hu-PXB and SCID mice. Data are shown as means ± S.D. of 5 mice.


Fig. 3. (A) Relationship between in vitro and in vivo human CLh,int. The solid line through origin represents nonlinear regression. The bold dashed line represents unity. The dashed lines represent twofold or 1/2-fold the unity and the dotted lines represent threefold or 1/3-fold the unity. The in vitro CLh,int of zidovudine was not calculated because its disappearance in the metabolic assay could not be observed. (B-C) Correlation in CLh,int. Correlation in CLh,int between SCID mice and humans (B), and between hu-PXB mice and humans (C). The solid lines through origin represent nonlinear regression. The dashed lines represent twofold or 1/2-fold the regression value and the dotted lines represent threefold or 1/3-fold the regression value. (D) Relationship of %change in CLtotal of lorazepam inhibited by probenecid (100 mg⋅kg−1) between Hu-PXB mice and humans.