Choosing the Best Tools for the Job – How to Improve DILI Prediction using a Multi-parametric Approach

As conferences across the globe are cancelled and postponed due to the COVID-19 outbreak, Cyprotex will be presenting its latest posters and presentations in digital format over the coming weeks and months.

Today, we take a look at the work behind one of our posters prepared for SOT 2020: Combined High Content Imaging in Liver Microtissues and Mitochondrial Toxicity to Predict Drug-induced Liver Injury (DILI).

Traditionally DILI has been difficult to predict during the discovery and pre-clinical development process. Often this leads to late stage failures during clinical development and in certain circumstances, can result in withdrawals following market approvals (e.g. troglitazone and cerivastatin).

With DILI being a common reason for drug attrition and drug withdrawals, it is apparent that traditional preclinical models are not sufficient to accurately de-risk drug-induced toxicity in humans.  Cyprotex is at the forefront of developing new combined strategies by building on existing single end-point 3D models.  In this research poster we examine the predictive capabilities of combining HepaRG spheroids using multi-parametric confocal High Content Screening (HCS) combined with endpoints from mitochondrial health assessment using the mitochondrial stress test, to predict DILI.

We also analyse the use of hLiMTs and HepaRG spheroid models – both models were individually cultured before cytochrome P450 (CYP) activity was determined.  After dosing with a set of known DILI-positive and negative compounds, a broad range of high content imaging (HCI) endpoints was assessed (in addition to cellular ATP) including:

  • DNA structure
  • GSH content
  • ROS formation
  • Mitochondrial function

The mitochondrial stress testing was performed using 2D HepG2 cells via the analysis of OCR (oxygen consumption rate), ECAR (extra cellular acidification rate) and reserve capacity using an Agilent Seahorse XFe96 instrument.

By utilising multi-parametric HCI in addition to ATP, a significantly improved sensitivity was observed over ATP content alone while specificity remained high.  Furthermore, the HepaRG spheroids appear to be comparable in sensitivity to hLiMTs in the prediction of DILI. This is likely related to the high level of CYP activity in both models.

A combined approach utilising multi-parametric HCI (in 3D HepaRG spheroids) and the mitochondrial stress test (in 2D HepG2 cells) correctly identified 80 out of 93 compounds as DILI-positive or negative, with only 4 false positive and 9 false negative compounds.

In conclusion, liver organoids subjected to a broad HCI endpoint strategy along with 2D cell-based mitochondrial stress testing represent a valuable addition to the in vitro DILI de-risking toolbox and should be considered alongside current methods.

Request a copy of this poster by contacting


This website is intended to assist industry participants, customers and employees to understand Cyprotex's global operations and ambitions.