Accurate High-Throughput Prediction of Hepatocyte Toxicity in Humans

Predicting internal organ toxicity with cell-based methods is challenging. We have developed high-throughput methods that predict nephrotoxicity [1] and hepatocyte toxicity in humans with high accuracy [2]. The methods use high-content imaging in combination with phenotypic profiling and machine learning. The method for the prediction of hepatocyte toxicity in humans employs HepaRG™ cells, which are currently one of the best cell models for addressing hepatocyte toxicity. The method has been validated with 69 reference chemicals known to be toxic / not toxic for hepatocytes in humans, such as drugs, industrial chemicals, and environmental toxicants. Whereas the sensitivity of current cell-based methods for hepatotoxicity prediction is ~50%-65%, our HepaRG-based method has a test sensitivity of 73% at a test specificity of 92%. Hepatocyte toxicity of large numbers of chemicals can be accurately predicted with this method, also in the absence of mechanistic data or other prior characterization. 

[1] R. Su, S. Xiong, D. Zink, L.H. Loo, Arch Toxicol 90(11) (2016) 2793-2808.

[2] F. Hussain, S. Basu, J.J.H. Heng, L.H. Loo, D. Zink, Arch Toxicol 94(8) (2020) 2749-2767.

Speaker: Daniele Zink

Daniele Zink is Team Leader and Principal Research Scientist at the NanoBio Lab (Agency for Science, Technology and Research, Singapore). Her work has received various international awards, including awards from the US Society of Toxicology and the prestigious LUSH Science Prize. Zink is co-founder of the spin-off Cellbae that offers stem cell and other products and safety screening services. She holds 12 patents/patent applications and has > 70 publications, which include publications in Nature, Nature Reviews Cancer, and Archives of Toxicology.