A Six-Stage Workflow for Robust Application of Systems Pharmacology
Kapil Gadkar, Dan Kiroauc, Don Mager, Piet van der Graaf, Saroja Ramanujan
Genentech, University at Buffalo, Leiden University
Objectives: The area of Quantitative Systems Pharmacology (QSP) is seeing increasing adoption and efforts in pharmaceutical research and industry settings. QSP however represents a different approach than more traditional pharmacometrics modeling and simulation, and thus, involves different technical considerations. QSP models are typically less driven by fitting of individual datasets but involve integration of diverse datasets to enable the mathematical representation of the biology of interest. Along with an expanded scope due to broader intended applications, this can potentially lead to under-constrained models. Further, QSP models are often used for testing biological/clinical hypotheses and for predictions in scenarios or patient populations where clinical data is limited, These novel aspects of QSP necessitate different technical workflows and approaches.
Methods & Results: Here we present a robust workflow that, in its entirety or in sections, has been successfully applied in QSP-based efforts to address many of the novel challenges these efforts face. This workflow involves: (1) initial data evaluation and scope specification; (2) model structure identification and implementation; (3) initial calibration & validation of “reference” virtual subjects and (4) of alternate virtual subjects and virtual “populations”; (5) model-based prediction; and (6) iteration with laboratory and clinical data acquisition. Technical approaches for each of these stages are discussed, including: aggregation of diverse data; selection of modeling formalism; development and identifiability of model structure(s); parameter optimization, sensitivity, and uncertainty/variability; resulting robustness of associated predictions; and experimental design guidance. In this context, we also review published systems modeling efforts that illustrate this workflow and the technical approaches discussed.
Conclusions: We have presented a staged workflow for the application of QSP. Notably, this workflow helps address several questions and criticisms commonly facing QSP projects. By providing a common, organized strategy along with guidance on technical approaches to address these and other considerations, we believe this workflow and subsequent evolution thereof can offer a useful framework for the execution, communication, and acceptance of QSP endeavors.