A framework to assess the translation of safety pharmacology data to humans.

ABSTRACT

This article outlines a strategy for collecting accurate data for the determination of the sensitivity, specificity and predictive value of safety pharmacology models. This entails performing a retrospective analysis on commonly used safety pharmacology endpoints and an objective assessment of new non-clinical models. Such assessments require a systematic quantitative analysis of safety pharmacology parameters as well as clinical Phase I adverse events. Once the sensitivity, specificity and predictive capacity of models have been determined, they can be aligned within specific phases of the drug discovery and development pipeline for maximal impact, or removed from the screening cascade altogether. Furthermore, data will contribute to evidence-based decision-making based on the knowledge of the model sensitivity and specificity. This strategy should therefore contribute to the reduction of candidate drug attrition and a more appropriate use of animals. More data are needed to increase the power of analysis and enable more accurate comparisons of models e.g. pharmacokinetic/phamacodynamic (PK/PD) relationships as well as non-clinical and clinical outcomes for determining concordance. This task requires the collaboration and agreement of pharmaceutical companies to share data anonymously on proprietary and candidate drugs.