Background: In recent years, Quantitative system pharmacology (QSP) modeling has gained significant attention as a valuable tool in drug development. The results of QSP modeling analysis have been used in different stages of drug development and submitted to Office of Clinical Pharmacology (OCP) to support decision-making. This assessment discusses challenges arising from validating complex and multiscale models and summarizes how a risk-informed credibility assessment framework was applied for evaluating QSP modeling in select regulatory submissions. Methods: QSP models have been developed and used in all stages of drug development to inform, for example, the selection of dose regimen and trial design, and support safety assessments. This assessment provides an overview of QSP modeling and analyses submitted to Center for Drug Evaluation and Research from 2021-2023 and discusses key regulatory considerations from the review of select QSP models. V&V40 standard is an FDA-recognized standard that provides a risk-based framework for establishing the credibility requirements of a computational model. Through case studies, we will walk through the concept of model risk assessment, and the role of regulatory scientists in evaluating QSP models and expectiations/limitions of QSP analysis in regulatory decisions. Results: Three case studies are summarized and presented to demonstrate the considerations and practice when reviewing a QSP model (e.g., assessments on model structure, model assumptions, biomarker selection, parameter calibration and optimization, virtual population generation, model validation and model application) for regulatory purpose. Conclusion: The V&V40 model credibility framework allows reviewers to effectively review submitted QSP analyses. Within the V&V40 framework, credibility assessments on the major QSP components (i.e., model structure and parameters, model validation and virtual population) depend on evaluation of model risk (model influence, decision consequence and overall model risk) for individual cases. In addition, reviewers are still accumulating experience to establish a good review practice for QSP modeling.
