The Role of High Content Toxicology and In Silico Modelling in Identifying Toxic Liabilities

The manifestation of in vivo xenobiotic-induced toxicity in an organism is multi-factorial. It depends on the potential of a xenobiotic to induce cellular or extra-cellular injury, the exposure (both in terms of concentration and duration) of susceptible sites in the organism to the xenobiotic, and potential modulatory effects dependent on the physiological, hormonal and dietary status of the exposed organism. In developing methods for predicting in vivo toxicity that are suitable for deployment in a pharmaceutical or chemical screening pipeline, appropriate attention must be paid to each of these factors.

High Content Screening (HCS) is a powerful and versatile multi-parametric tool for quantifying and understanding the changes that occur in cells when they are exposed to xenobiotics. By selecting appropriate end-points (e.g. impact on mitochondrial function, induction of stress response, cell death) in a high content screen, the toxic potential of a xenobiotic can be measured and quantified.

The exposure of internal susceptible sites in populations that are exposed to a xenobiotic (e.g. patients taking drugs; animals used in toxicity testing; consumers and workers exposed to environment-contaminating chemicals) can be predicted from appropriate in vitro compound data by techniques such as physiologically based pharmacokinetic (PBPK) modelling. This technique is also suitable for modelling the dynamic response of exposed cells and the modulating effects of the organism’s physiological status.

We describe results obtained with the CellCiphr system, which combines high content screening with an innovative ranking and classifying system for predicting in vivo toxicity. CellCiphr screening data are generated in multiple cell lines (primary rat hepatocytes, HepG2 human liver cells and H9c2 cardiomyocytes), at multiple time points across multiple endpoints. The endpoints probe the interacting network of processes that are involved in cellular function and cellular toxic response, and have been chosen for their relevance in predicting aspects of drug-induced injury. The large amount of data generated for each compound provides a uniquely detailed series of snapshots of responses to toxic challenge. It lays the foundation for understanding the mechanisms of toxicity induced by any given compound, and the potential for different organs to undergo different responses.

A unique component of CellCiphr is a set of proprietary in silico models for predicting the risk of compound failure in safety studies. These models combine the detailed CellCiphr results for a test compound into a single toxicity ranking, which allows compound prioritization based on in vitro toxicity. These models have been built with our extensive database of CellCiphr results across multiple cell panels. We present and discuss relative toxicity rankings of a selection of results for HepG2 cells and rat hepatocytes, and also explore relationships between different CellCiphr endpoints. We also discuss the prediction of internal exposure using PBPK modeling, and the role that such modeling will play in the further development of robust in vivo toxicity prediction from HCS, and other, in vitro toxicity data.