Session 3: Using in vitro Enzyme and Transporter Data in Translational Models of Human Pharmacokinetics, Dose and DDI
Chairs: Jashvant Unadkat, University of Washington, Seattle, Washington, USA and Kimio Tohyama, Takeda Pharmaceuticals International Co., Cambridge, Massachusetts, USA
Translation of in vitro and in silico ADME Data to Guide Drug Discovery and Development
Marcel Hop, Genentech Inc., South San Francisco, California, USA
PBPK Modelling and Simulation of Transporter-mediated PK and DDIs – Current Status and Challenges 
Aleksandra Galetin, University of Manchester, Manchester, England, United Kingdom
Application of PBPK using a Matrix Qualification Approach for Translational Predictions of OAT1/OAT3 Inhibition in the Clinic by Cabotegravir 
Kunal Taskar, GSK, United Kingdom
IVIVE of Transporter-Mediated Renal Clearance: Relative Expression Factor (REF) vs Relative Activity Factor (RAF) Approach
Aditya Kumar, Department of Pharmaceutics, University of Washington, Seattle, Washington, USA
Successes and Challenges Faced by Industry in IVIVE of Transporter-based Drug Disposition and DDIs
Manthena Varma, Pfizer, Niantic, Connecticut, USA
IVIVE of Biliary Clearance of Rosuvastatin: a Comparison of the Proteomics-informed REF Approach vs. Sandwich-Cultured Human Hepatocytes (SCHH)
Flavia Storelli, University of Washington, Seattle, Washington
Validation of PXB Chimeric Mice to Predict Human Liver-to-Plasma Kpuu of OATP1B1 Substrates
Bo Feng, Vertex Pharmaceuticals, Boston, Massachusetts, United States
Panel Discussion with All Speakers
Key:
Complete
Next
Failed
Available
Locked
Translation of in vitro and in silico ADME Data to Guide Drug Discovery and Development | Dr. Marcel Hop
Recorded 03/04/2021
Recorded 03/04/2021
Based on the number of FDA approvals it appears that the productivity of the pharmaceutical industry is improving. In addition, the number of drugs in development that have failed because of poor pharmacokinetics has steadily declined. Thanks to better in vitro ADME models and sophisticated PBPK modelling and simulation we are able to predict the human PK reliably in most cases. However, the drug discovery process remains very inefficient. Frequently, several thousand compounds are synthesized in the course of a project and multi-parameter optimization is a trial and error process. Indeed, identifying a compound with the right balance of properties is still a lengthy process. One of the reasons for this is that ADME scientists frequently do not speak the same language as medicinal chemists and, therefore, are not as influential in design of new molecules as desired. Fortunately, the large quantity of in vitro ADME data gathered over the last 10-20 years has enabled the creation of in silico models that can be used to predict ADME properties and triage design ideas prior to synthesis of the compounds. In silico ADME models are available to predict in vitro ADME properties such as metabolic stability and sites of metabolism, permeability and efflux, plasma protein binding, and CYP inhibition and ultimately in vivo PK thereby reducing the time required to find compounds with the right balance of properties. Subsequently, models are available that incorporate a mix of predicted and measured properties (e.g., potency) to prioritize compounds efficiently for further optimization. In this presentation, the author will discuss the value of these models and how to incorporate them effectively in lead optimization to enable rapid idea generation and ultimately optimization. Getting the expectations right and shifting the mindset to one based on probability of success is essential for successful implantation of these in silico models. The author will also speculate about the future including one where AI/ML will be used to predict potency allowing complete integration of properties and, hence, prioritization.
PBPK Modelling and Simulation of Transporter-mediated PK and DDIs – Current Status and Challenges | Dr. Aleksandra Galetin
Recorded 03/04/2021
Recorded 03/04/2021
Transporter-mediated pharmacokinetic and drug-drug interactions (DDIs) often involve interplay between multiple transporters and/or enzymes in different tissues. In many cases these complex DDIs result in differential effect on systemic and tissue drug exposure which may impact either drug safety and/or efficacy. Physiologically-based pharmacokinetic (PBPK) modelling has increasingly been used as a translational tool to gain mechanistic insight into interplay of these multiple processes and to support regulatory submission, in particular for DDIs involving hepatic transporters and transporter-enzyme interplay. In contrast, examples of mechanistic PBPK modelling of transporter-mediated disposition in other organs (e.g., kidney) are less evident. This presentation will focus on examples of PBPK best practices and challenges with respect to verification of PBPK model-predicted changes in tissue exposure. Appropriate use of clinical data for reverse translation will be discussed. In addition, PBPK modelling of special populations will be illustrated with example of modelling of impaired renal function. Current challenges to delineate the effect of disease and renal transporter inhibition in this patient population will be presented using creatinine example.
Application of PBPK using a Matrix Qualification Approach for Translational Predictions of OAT1/OAT3 Inhibition in the Clinic by Cabotegravir | Dr. Kunal Taskar
Recorded 03/04/2021
Recorded 03/04/2021
Cabotegravir (CAB) is an integrase strand transfer inhibitor being investigated for the treatment and prevention of HIV-infection. It is being developed as a long acting (LA) intra-muscular injection to facilitate every 1 or 2-month dosing. It is necessary to evaluate the impact of CAB on the exposure and clearance of co-medications. In vitro studies indicated that CAB inhibits renal transporters OAT1 and OAT3 with half maximal inhibitory concentrations of 0.81 and 0.41 µM, respectively. The objective of the present analysis was to build a physiologically based pharmacokinetic (PBPK) model of CAB to predict the clinical implications of renal OAT1/OAT3 inhibition on co-medications. Further, suitable qualification and verification of the PBPK platform and compound files is necessary to gain confidence in the PBPK predictions as well as to comply with the regulatory expectations. Methods: A mechanistic PBPK model of CAB in the adult population was built using the Simcyp® v17.1 simulator by incorporating physico-chemical properties, in vitro clearance mechanisms, and in vivo data and validated as per regulatory specifications. The CAB PBPK model was validated through comparison with available clinical PK data following oral CAB 30mg administration in healthy volunteers and qualified against predicting OAT1 and/or OAT3 inhibition based DDIs. DDI simulations were performed to evaluate the effect of CAB oral doses on the exposure of OAT1/OAT3 substrates (methotrexate, tenofovir, ciprofloxacin, cidofovir, cefuroxime, oseltamivir carboxylate, baricitinib, and S44121). A matrix approach was used for the qualification of the platform as well as the probes substrates and inhibitors (probenecid, diclofenac) used in this analysis.
Results: Simulated DDIs for above mentioned OAT1/OAT3 substrates and inhibitors were within two-fold of the observed clinical DDIs. This qualified the Simcyp® v17.1 simulator and related files as appropriately sensitive for predicting OAT1/OAT3 inhibition-mediated clinical DDIs. CAB PBPK model accurately predicted CAB PK parameters (all within acceptable bioequivalence criteria (0.80-1.25) for single as well as repeat dose studies). DDI simulations predicted a mean change in systemic exposure for tested OAT1/OAT3 substrates of <25% after co-administration with CAB at steady state. Conclusions: A PBPK model of CAB was developed and validated that accurately predicted human pharmacokinetics observed in healthy volunteers. A matrix approach helped in the platform as well as compound file qualification to gain confidence in the clinical predictions with the PBPK analysis. OAT1/OAT3 substrate drugs such as tenofovir, cidofovir, methotrexate were predicted to have a minimal risk of DDIs when administered with CAB. Similar CAB concentrations following oral and LA administration suggest that these results would apply to CAB LA. The predicted lack of interactions supports co-administration with OAT1/OAT3 substrates without dose adjustments.
IVIVE of Transporter-Mediated Renal Clearance: Relative Expression Factor (REF) vs Relative Activity Factor (RAF) Approach | Dr. Aditya Kumar
Recorded 03/04/2021
Recorded 03/04/2021
A comprehensive analysis of 391 drugs found that 31% of drugs are predominantly renally cleared (CLr) (i.e. CLr > 50% of total clearance). Of these, many are actively secreted by transporters. For these drugs, in vitro-to-in vivo extrapolation (IVIVE) of transporter-mediated renal secretion is important to prospectively predict their in vivo renal secretory CL (CLsec,plasma) as well as CLr,plasma and to assess the impact of drug-drug interactions and pharmacogenetics on their pharmacokinetics. In this study we compared the ability of the relative expression factor (REF) and the relative activity factor (RAF) approaches to quantitatively predict the in vivo CLsec,plasma and CLr,plasma of 26 Organic Anion Transporter (OAT) substrates from in vitro studies using transporter-expressing cells (TEC). Briefly, the REF approach requires measurement of the abundance of each transporter of interest in the relevant tissue (in this case the kidneys) and in vitro(in TEC). Then, the ratio of these abundances is used to scale the in vitro transporter-mediated drug clearance to that in vivo. In contrast, the RAF approach requires measurement of in vivo and in vitrotransporter-mediated clearance of a transporter-selective probe substrate (for each transporter of interest). Then, the ratio of these clearances is used to scale the in vitro transporter-mediated drug clearance to that in vivo. We found that for the REF approach, 50% and 69% of the CLsec,plasma predictions were within 2- and 3-fold of the observed in vivo values, respectively; while the corresponding values using the RAF approach were 65% and 81%, respectively. Also, for the REF approach, 65% and 92% of the CLr,plasma predictions were within 2- and 3-fold of the observed in vivo values, respectively; while the corresponding values using the RAF approach were 81% and 88%, respectively. Despite these numerical differences, the two approaches were not significantly different in their ability to predict (as measured by precision and bias) CLsec,plasma or CLr,plasma of the OAT drugs examined. Given these results, we recommend that the REF and RAF approaches can be used interchangeably when predicting OAT-mediated CLr. Funded in part by Pfizer Inc., UWRAPT and NIH GM007750.
Successes and Challenges Faced by Industry in IVIVE of Transporter-based Drug Disposition and DDIs | Dr. Manthena Varma
Recorded 03/04/2021
Recorded 03/04/2021
Active and/or passive drug transport in the liver governs the drug availability for subsequent biotransformation by the drug-metabolizing enzymes or efflux across the canalicular membrane into bile. This ‘transporter-enzyme interplay’ is best described by extended clearance concept. A framework, Extended clearance classification system (ECCS), based on simple drug properties was proposed to enable prediction of the predominant clearance mechanism, and potentially identify the prominent transporter activity. Several solute carriers (SLCs) including OATP1B1/1B3/2B1, NTCP, OAT2 and OCT1 are highly expressed on the sinusoidal membrane and have the potential to dictate hepatic clearance. Characterizing the contributions of one or multiples of SLCs and passive uptake clearances is critical in prediction/evaluation of pharmacokinetics and PK variability due to drug-drug interactions, pharmacogenomics and disease state. Additionally, building robust IVIVE approaches is critical for the quantitative PK predictions. Several in vitro tools (eg. cell cultures, vesicles, primary cells, etc.) are available to characterize transport mechanisms and estimate relevant kinetic parameters. This presentation will discuss utility of in vitro and in vivo models to enable hepatic uptake clearance characterization (transporter phenotype) with case examples; and review current drug discovery IVIVE strategies to predict pharmacokinetics and drug interactions. Emphasis will be placed on the current gaps along with the emerging approaches to gain further confidence in the IVIVE of transporter clearance.
IVIVE of Biliary Clearance of Rosuvastatin: a Comparison of the Proteomics-informed REF Approach vs. Sandwich-Cultured Human Hepatocytes (SCHH) | Dr. Flavia Storelli
Recorded 03/04/2021
Recorded 03/04/2021
Purpose: Extrapolating in vivo human biliary CL (CLbile) of drugs from in vitro data (i.e. IVIVE) is important to predict the impact of biliary transporter-based drug interactions or pharmacogenetics on systemic and hepatic drug concentrations. Predicting hepatic drug concentrations is particularly important when the liver is the site of drug efficacy or toxicity. Currently, sandwich-cultured human hepatocytes (SCHH) are routinely used to predict CLbile, but their use is limited by their cost, low throughput, need for transporter-qualified SCHH, uncertainty in their ability to obtain an estimate of CLbile and, when estimated, the potential to overpredict CLbile [1]. Therefore, we determined, using rosuvastatin (RSV) as a model drug, if the relative expression factor (REF) approach could be a more reliable, higher throughput and accurate method to predict CLbile of drugs. Our recently determined rosuvastatin (RSV) CLbile, obtained using Positron Emission Tomography (PET), was used for verification of the RSV CLbile prediction by either the REF or the SCHH approach [2]. Method: We determined the transporter-based intrinsic CL of RSV in BCRP-, P-gp- and MRP2- overexpressing vesicles and scaled these CLs to in vivo using REF, the ratio of transporter abundance in human liver tissue [3-5] and transporter-expressing vesicles (measured by targeted proteomics). REF was corrected for the percentage of inside-out vesicles measured by the ATPase or 5’-nucleotidase assays or obtained from the literature for vesicles from the same supplier. RSV CLbile was estimated by compartmental modeling of human SCHH (4 different donors) uptake and efflux data obtained in the absence or in presence of calcium (as previously described by us [6]). The vitro data CLbile was scaled to in vivo based on total protein content in liver tissue vs. in vitro. Then, the predicted in vivo CLbile, using the two approaches, was compared with our in-house human [11C]RSV PET imaging data [2]. Results: RSV was found to be a substrate of BCRP, P-gp and MRP2 in transporter-overexpressing vesicles. After scaling each individual CLs to in vivo using REF, we found that the contribution of BCRP, P-gp and MRP2 to RSV CLbile was about 51%, 9% and 40%, respectively. REF-predicted (6.1 ml/min) and SCHH-predicted (65.4 ml/min) RSV CLbile were respectively 1.20-fold and12.9-fold of the CLbile value reported by our PET imaging study [2]. The overprediction of CLbile by SCHH was reduced to around 2-fold after correcting for the overexpression (total abundance) of biliary efflux transporters in human SCHH vs. liver tissue as determined by proteomics [7]. Conclusion: Using the REF approach, we successfully predicted the in vivo PET-imaged RSV CLbile mediated primarily by BCRP and MRP2. As expected, the SCHH overpredicted CLbile of RSV and this prediction was closer to the in vivo CLbile only when corrected for the overexpression of biliary transporters in SCHH vs. liver tissue. This study demonstrates the success of the REF approach in predicting in vivo biliary CL which, in combination with bi-directional sinusoidal CL and intracellular metabolic CL of the drug, can predict intracellular hepatic drug concentrations. While the REF approach is high-throughput, cost-effective and accurately predicted the in vivo CLbile, of RSV, PET-imaged CLbile estimates of additional drugs are needed to support this conclusion.
Acknowledgement: This work was supported in part by the UWRAPT through funding from Gilead Sciences, Amgen, Biogen, Genentech, Merck, and Takeda.
Validation of PXB Chimeric Mice to Predict Human Liver-to-Plasma Kpuu of OATP1B1 Substrates| Dr. Bo Feng
Recorded 03/04/2021
Recorded 03/04/2021
The ability to predict human liver-to-plasma unbound partition coefficient (Kpuu) is important to estimate unbound liver concentration for drugs that are substrates of hepatic organic anion transporting peptide (OATP) transporters with asymmetric distribution into the liver relative to plasma. Herein, we explored the utility of PXB chimeric mice with humanized liver that are highly repopulated with human hepatocytes to predict human hepatic disposition of OATPs substrates, including rosuvastatin, pravastatin, pitavastatin, valsartan and repaglinide. In vitro total uptake clearance and transporter-mediated active uptake clearance in C57 mouse hepatocytes were greater than PXB chimeric mouse hepatocytes for rosuvastatin, pravastatin, pitavastatin and valsartan. Consistent with in vitro uptake data, enhanced hepatic uptake and resulting total systemic clearance were observed with the above four compounds in control SCID than in PXB chimeric mouse, which suggest that mouse has a stronger transporter-mediated hepatic uptake than human. In vivo liver-to-plasma Kpuu from PXB chimeric and SCID control mice were also compared, and rosuvastatin and pravastatin Kpuu in SCID mouse were more than 10-fold higher than that in PXB chimeric mouse, whereas, pitavastatin, valsartan and repaglinide Kpuu in SCID mouse were comparable with Kpuu in PXB chimeric mouse. Finally, PXB chimeric mouse liver-to-plasma Kpuu values were compared with the reported human Kpuu, and a good correlation was observed as the PXB Kpuu vales were within 3-fold of human Kpuu. Our results indicate that PXB mice could be a useful tool to delineate hepatic uptake and enable prediction of human liver-to-plasma Kpuu of hepatic uptake transporter substrates.