Can ToxProfiler guide DILI predictions?

Background

The need for human relevant models to predict DILI

A common aim of food, agrochemical, cosmetic, and pharmaceutical industries is to identify early in the development process whether candidate compounds have the potential to cause liver toxicity and Drug-Induced Liver Injury (DILI). This goal is often hampered by the fact that there can be multiple mechanisms occurring, either alone or in combination, which manifest as liver toxicity. Animal-based studies are not always the answer, since even though they represent an intact model (and taking aside the high cost in terms of animals, time and budget), distinct species differences in metabolism and physiological mechanisms mean that liver toxicity occurring in humans are often not detected in animals.

Clearly, human-relevant models are needed to predict the potency of novel pharmaceuticals to induce liver toxicity, which means researchers must detect DILI effects using in vitro human models. Endpoints can range from simply measuring cytotoxicity, which does not help with revealing how DILI occurs, to omics technology, which increases the biological coverage but opens a whole can of worms as to how to interpret an immense amount of data from non-standardized incubations, gene/protein/metabolite measurement methods, and biostatistical analyses (all of which are time- and budget-consuming).

The need

Simply identifying a compound as “DILI”-positive only provides an indication of hazard; however, current safety strategies are moving towards identifying the toxicological mode of action and exposure-related assessments. A margin of safety is derived from a comparison of the point at which toxicity/bioactivity is first observed, the point of departure, and plasma concentrations after exposure to the compound. Therefore, any approach aiming to predict DILI potential should do so by covering relevant pathways, as well as providing a quantitative measure of the effect to derive a point of departure. An ideal screening assay panel would thus be:

• Human-relevant
• Applicable to early development (higher throughput, reproducible, simple and sensitive readout)
• Covering biological pathways relevant to DILI
• Quantitative, allowing the derivation of a point of departure
• Able to reliably identify known DILI compounds.

The solution

Although not originally designed to specifically address DILI, the ToxProfiler reporter assay checks all but the last of these attributes – until now. ToxProfiler is a panel of human reporter in vitro assays conducted in 384-well format measuring the activation of 7 key cellular stress response pathways involved in different toxicological effects. Compounds acting via these pathways can be detected with a high sensitivity at the single cell level using a simple readout of live cell confocal imaging of these fluorescent markers. Quantification of the concentration-dependent responses enables the derivation of a point of departure and the primary toxicological mode of action e.g., via oxidative stress (SRXN1-GFP), cell cycle stress (p21-GFP), endoplasmic reticulum (ER) stress (CHOP-GFP), ion stress (MT1X-GFP), protein stress (HSPA1B-GFP), autophagy (LC3-GFP) and inflammation (ICAM1-GFP). ToxProfiler can also be adapted to include metabolic supplementation to ascertain whether the parent or a metabolite is responsible for DILI.

Our Approach

Initial data supporting the final attribute have recently been generated, whereby over 200 compounds with known clinical and non-clinical toxicity profiles were screened in ToxProfiler. The ToxProfiler panel was able to identify known toxicological mode of action of a diverse set of compounds at non- or sub-cytotoxic concentrations. When the point of departures of the compounds were ranked in addition to clustering, ToxProfiler correctly predicted the primary toxicological mode of action for 90% of the compounds, including those which cause DILI. Results of the assays were visualized using hierarchical clustering to group chemicals based on their ToxProfiler stress reporter signatures. This resulted in different clusters with several of the most enriched clusters linked to specific mechanisms, and even to specific pathologies, including DILI. Indeed, the ToxProfiler reporter panel identified oxidative stress, ER stress, and autophagy as the most important predictors of DILI, which is in line with the findings of others (Wink et al., 2017, Wu et al., 2023).

ToxProfiler in DILI Prediction

Of the 200 compounds tested in a large ToxProfiler validation study, the FDA DILI liabilities of 64 were known (Chen et al. 2016). These were classified as compound with “most DILI concern”, “less DILI concern” or “no DILI concern”. To classify the compounds for their DILI liabilities in ToxProfiler, the ratio between bioactivity and the clinical plasma concentration was calculated. Using this bioactivity exposure ratio, ToxProfiler correctly classified 30/31 compounds with “most DILI concern” liabilities and 6/7 with “no DILI concern” resulting in sensitivity and specificity scores of 97% and 86% respectively. Although the 26 “less DILI concern” compounds also had a higher bioactivity exposure ratio in ToxProfiler, it was more difficult to distinguish the “less DILI concern” from the “no DILI concern” compounds.

Conclusion

Clearly, further investigations are needed to extend the number of compounds tested. However, these initial results point to ToxProfiler being a promising tool for early safety assessments – checking all relevant attributes for an ideal screening assay for this pathology. We will keep you posted on our progress!

Download the case study

Request presentation for details

Chen M, Suzuki A, Thakkar S, Yu K, Hu C, Tong W. DILIrank: the largest reference drug list ranked by the risk for developing drug-induced liver injury in humans. Drug Discov Today. 2016 Apr;21(4):648-53. doi: 10.1016/j.drudis.2016.02.015. Epub 2016 Mar 3. PMID: 26948801.

Wink S, Hiemstra S, Herpers B, van de Water B. High-content imaging-based BAC-GFP toxicity pathway reporters to assess chemical adversity liabilities. Arch Toxicol. 2017 Mar;91(3):1367-1383. doi: 10.1007/s00204-016-1781-0. Epub 2016 Jun 29. PMID: 27358234; PMCID: PMC5316409.

Wu H, Bao X, Gutierrez AH, Nevzorova YA, Cubero FJ. Role of oxidative stress and endoplasmic reticulum stress in drug-induced liver injury. Explor Dig Dis. 2023;2:83–99. https://doi.org/10.37349/edd.2023.00020