Applications of Model-Based Target Pharmacology Assessment in Defining Drug Design and DMPK Strategies: GSK Experiences

Abstract

Many critical decisions faced in early discovery require a thorough understanding of the dynamic behavior of pharmacological pathways following target engagement. From fundamental decisions on the optimal target to pursue and the ultimate drug product profile (combination of modality, potency, and compound properties) expected to elicit the desired clinical outcome to tactical program decisions such as what chemical series to pursue, what chemical properties require optimization, and what compounds to synthesize and progress, all demand detailed consideration of pharmacodynamics. Model-based target pharmacology assessment (mTPA) is a computational approach centered around large-scale virtual exploration of pharmacokinetic and pharmacodynamic models built early in discovery to guide these decisions. The present work summarizes several examples (use cases) from programs at GlaxoSmithKline that demonstrate the utility of mTPA throughout the drug discovery lifecycle.

Model-Based Virtual PK/PD Exploration and Machine Learning Approach to Define PK Drivers in Early Drug Discovery

“…24 The ADME properties and potency of each virtual compound, predicted using in silico models, were then used as input to a PBPK/PD model to determine the minimum effective dose (MED) as previously described. 3 Machine learning models were then built using molecular fingerprints to fit the calculated MEDs. SHAP analysis was then used to rank order structural features (instead of PK end points) according to their importance in impacting MED.…”

“…Not only could this greatly reduce the cycle time to achieve a clinical candidate, but the prospective candidate should carry a higher probability of success in clinical trials because the appropriate PK and PD drivers were considered from the outset of the project. Several real-life examples of how such an approach was used to guide target selection, definition of screening and progression criteria, and prioritization of compound series have been reported previously in great detail …”

“…If a small variation in the model leads to significant changes in the rank order of structural features, then one could opt to proceed with parallel optimization efforts based on different sets of model parameters or initiate experimentation to better define the parameter earlier. Another option is to conduct tSNE cluster mapping, as previously described, using different versions of the model or different sets of model parameters to determine if virtual compounds predicted to be effective cluster around the same compound space or not. More generally, one could adapt an iterative exploit-versus-explore approach to compound optimization in the face of uncertainty in model predictions .…”

“…Many of the computational methods for mTPA have been previously described , and will only be briefly recited in addition to any extension to the methods.…”

“…Detailed examples applying this methodology to guide medicinal chemistry and DMPK strategies have been published previously. 3 The virtual exploration of PK/PD relationships unique to each pharmacological process of interest used in this study is summarized in Figure 1. Details of this method have been previously described as part of the model-based target pharmacology assessment (mTPA) methodology, 1,3 and is hereafter referred to as mTPA analysis.…”