Issue link: https://www.e-digitaleditions.com/i/1354372
www.biopharminternational.com March 2021 eBook BioPharm International 29 the observed DDI. Mechanisms of drug interaction are also import- ant; for instance, the multiple interaction mechanisms (CYP2C8 and OATP1B1) of both parent and metabolite of gemf ibrozil were simultaneously implemented in PBPK models to predict DDIs of representative victim drugs. While consideration of parent alone led to a significant underprediction of the DDI, the inclusion of gemfi- brozil-glucuronide metabolite in the PBPK model greatly improved the acc urac y of predicted area under the curve (AUC) ratios and plasma concentration–time profiles of victim drugs. The gemfibrozil case study reinforces the utility of in-vitro data and the modeling approaches that mechanistically integrate the multiple components in the DDI risk assessment. I ncor p orat ion of met ab ol ite DDI within PBPK models was also impor tant for the itracona zole (ITZ) PBPK model built in collabo- ration with IQ member companies in response to the FDA restriction of the CYP3A4 strong inhibitor, ketoconazole, for use in clinical DDI studies (13). The goal of this working group was to develop and ver if y a mechanistic I TZ PBPK model that could be used to predict the impact of this strong CYP3A4 inhibitor on CY P3A subst rates (e.g., new drug candidates) with high confidence. Additionally, the model was used to provide recom- mendations on optimal DDI study design based on PBPK model sim- ulations. Using in-vitro and clinical PK data for ITZ and its metabolites collected from IQ member compa- nies, the PBPK model developed in this study was built mostly 'bot- tom-up' and was able to describe the PK accumulation of both ITZ and its major CYP3A4-inhibitory metabolite, hydroxy-ITZ (OH-ITZ), after repeat dosing of ITZ as a solu- tion and capsule under both fasted and fed states. Overall, the model predicted the DDI of I TZ w ith seven different CYP3A substrates (24 AUC ratios) accurately, thus ver- ifying the model for prospective use in DDI predictions. This verified model was then used to simulate various clinical DDI study scenarios considering formulation, duration of dosing, dose regimen, and food status to recommend the optimal design for maximal inhibitory effect by ITZ. For instance, recommenda- tions to maximize the DDI effect included loading dose suggestions, use of solution versus capsule, and length of ITZ dosing after substrate administration (e.g., continuing through 4–5 substrate half-lives). Due to the increasing use of ITZ as a tool in clinical DDI studies across pharmaceutical companies, the development of this mechanistic PBPK model for the accurate pre- diction of CYP3A4-mediated DDIs was an important collaboration among IQ companies and a benefit for use in clinical research and use- ful to industry and academia. Another area in which IQ eval- uated DDI modeling using PBPK was to assess the prediction of C Y P3A a nd C Y P2B6 induc t ion compared to other static model- ing approaches (9, 14). In these peer-reviewed publications, sev- eral DDI prediction models were evaluated for their ability to iden- tify drugs with CYP3A or CYP2B6 induction liabilit y based on in- vitro mRNA data. The drug inter- action magnitudes of CYP3A or CYP2B6 substrates were predicted using various static and mecha- nistic dynamic PBPK models. For C Y P3A, t he models per for med with high fidelity and predicted few false negatives or false posi- tives, with the basic models hav- ing a good predictive performance for identifying risk with respect to induction. For CYP2B6, the tri- als using the substrate efavirenz rather than bupropion were better predicted with the models. In gen- eral, it is agreed among IQ compa- nies that PBPK models are useful to optimally design DDI studies, pa r t ic u la rly t hose of comple x DDIs including dosing regimen, sampling strategy, study duration, and the strategy to differentiate the effects from induction and inhibition. Although there was a limited data set used in these stud- ies (e.g., more moderate to strong inducers of CYP3A, for instance), mechanistic models may be a com- plementary technique when a false positive result is suspected using basic models for induction. These case examples demonstrate the increased importance of establish- ing the use of PBPK modeling in DDI predictions through the IQ TALG over the past few years. IMPACT OF PBPK MODELING ON ASSESSING THE EFFECT OF ORGAN IMPAIRMENT ON HUMAN PHARMACOKINETICS The safe and effective use of drugs requires an understanding of how impaired renal and hepatic drug elimination impacts drug exposure in affected patients. Traditionally, the pharmaceutical industry has used stand-alone renal and hepatic impa ir ment st ud ies somet imes complemented with population PK modeling. These studies are not always available and of ten recruit small numbers of patients with diverse disease, and are not conducted until later in the drug development process. Predicting the outcome of such studies with PBPK models would be very valu- able, but the performance of cur- rently available organ impairment models lacks thorough evaluation. The IQ TALG Organ Impairment Regulatory Sourcebook Quality Collaboration