Drug induced liver injury (DILI) is one of the most challenging areas of toxicology to predict as preclinical toxicity testing often fails to pick up human adverse effects. Species differences in drug metabolism are thought, in part, to be responsible for this lack of translation. For this reason, human cell-based assays are now becoming a standard approach to assess potential DILI risk at an early stage. These cell‑based models need to be organ-specific and coupled with sensitive and specific endpoints which accurately predict drug-induced toxicity.
Since the introduction of next generation sequencing and high throughput RNA-seq, the value of transcriptomic analysis is now being realised within the pharmaceutical industry for safety assessment. Combining this with machine learning (ML) and artificial intelligence (AI) really helps to contextualise the data, allowing not just a safety prediction but also a mechanistic insight of adverse outcome pathways.
At the 61st Annual Society of Toxicology Meeting (SOT) and ToxExpo from March 27-31, 2022, we presented our research on ‘High throughput RNA-seq profiling of primary human hepatocytes and human liver microtissues to predict drug induced liver injury’. This study evaluated 128 reference compounds including 68 DILI-associated compounds and 60 non-DILI-associated compounds which were incubated with either primary human hepatocytes following a single dose over 24 hr or human liver microtissues with repeat dosing over 7 days. High throughput RNA-seq data (ScreenSeqTM) were generated and then processed and analysed using our multi-omic data analysis platform EVOpanHunter. Two approaches were used to predict DILI liability from the RNA-seq data including:
- benchmark dose accumulation curves
- ML/AI algorithms
Accuracy between the different cellular models and prediction models was compared. Specific case studies were used to demonstrate the advantages of the prediction models with ML/AI being the most robust cross-validation approach.
Find out more about our presence at SOT here.
To download the poster, click here.