Prediction of Liver Toxicity Revolutionized in Largest 3D In Vitro Benchmarking Study

The use of human liver microtissues (hLiMTs) cultured in a 384-well microplate system was explored as a predictive model for assessing drug-induced liver injury (DILI). This multi-institutional study, led by Bruno Filippi and Lola Fas at InSphero and published in the journal, Toxicological Sciences, evaluates 152 FDA-approved small molecule drugs, measuring cytotoxicity through ATP content changes to assess the model’s reliability for preclinical hepatotoxicity screening. With promising sensitivity and specificity for identifying high-risk DILI drugs, the study demonstrates hLiMTs as a scalable, physiologically relevant tool for early drug safety evaluation.

Enhancing drug safety assessment through improved liver models

There are limitations with current two dimensional (2D) in vitro models in predicting DILI, a major cause of drug withdrawals and failure in clinical trials. Traditional 2D liver cell cultures often lack physiological relevance, leading to a lack of predictivity for hepatotoxicity in humans. The current study sought to address these issues.

The three dimensional (3D) InSight™ Human Liver Microtissues are spheroid models comprising co-cultured primary human hepatocytes and nonparenchymal liver cells. By using physiologically relevant hLiMTs in a 384-well microplate format, we aimed to create a more scalable, robust and reproducible model that could reliably detect hepatotoxic drugs early in drug discovery. Testing a library of 152 Food and Drug Administration (FDA)-approved drugs, we evaluated the model's ability to detect high-risk DILI drugs, ultimately aiming to enhance drug safety assessments and reduce adverse events.

3D liver microtissues effectively predict hepatotoxicity

Using the 3D spheroid liver microtissue models, the hepatotoxicity of 152 FDA-approved small molecule drugs, stratified by DILI risk, was evaluated using adenosine triphosphate (ATP)-based cytotoxicity assays. The results were analyzed based on the drugs’ known hepatotoxicity classifications, drug labels and clinical exposure (Cmax).

The key findings of the paper were:

• The 3D InSight ™ hLiMT model correctly flagged 80% of the withdrawn drugs due to liver toxicity, while it accurately classified 89% of the safe drugs. For liver-toxic drugs targeting the nervous system, the liver microtissues achieved a 90% success rate.
• The use of 384-well Akura™ Spheroid Microplates enabled high-throughput screening, making the system scalable and ideal for large-scale hepatotoxicity assessment applications.
• Reproducibility and robustness: The model showed consistent results across different experimental runs and batches, supporting its reliability as an in vitro alternative for hepatotoxicity screening.
• C2C ratio as a discriminator: The IC50ATP-to-Cmax ratio (C2C) of 176 served as an effective threshold to distinguish hepatotoxic drugs (with a sensitivity of 71.7% and specificity of 88.9%), especially those labeled with severe warnings, such as "box warning" or "withdrawn."

A scalable, robust and reproducible drug screening system reduces attrition

The study demonstrates that 3D InSight™ hLiMTs in a 384-well microplate format offer a robust and reliable in vitro model for predicting DILI.

The study's findings support the model's utility as a high-throughput tool for preclinical hepatotoxicity screening, offering improved predictive accuracy and throughput compared to conventional systems. The hLiMTs’ ability to mimic in vivo liver morphology and functions closely, coupled with robust reproducibility across runs and batches, adds value to drug safety assessments. This 3D model could serve as a practical alternative to current screening methods, potentially refining early detection of hepatotoxic drugs and reducing the likelihood of liver-related adverse events in clinical trials and post-market stages.

The most important impact of this work and other studies like this is on patients and their families as the findings may lead to safer medications on the market, reducing the risk of drug-induced liver injury and adverse health outcomes. In addition to that, other groups also benefit from studies like these including:

Pharmaceutical companies: With more accurate preclinical hepatotoxicity screening, drug developers can identify potentially harmful drugs earlier, reducing late-stage trial failures and associated costs. Due to the scalability, the model could even be used in medicinal chemistry.

Regulatory agencies: Improved in vitro models can help regulatory bodies like the FDA ensure drug safety more effectively, supporting better guidelines for hepatotoxicity assessments.

Scientists: This work provides a new tool for studying drug toxicity mechanisms in a physiologically relevant model, enhancing the ability to generate reliable data.

Animal welfare advocates: The adoption of reliable 3D liver models reduces the need for animal testing in early toxicity screening, aligning with ethical considerations and reducing reliance on animal models.

While this study demonstrates promising results with hLiMTs for predicting hepatotoxicity, there are a few limitations. Although hLiMTs offer an improved physiological model over 2D cultures, they cannot fully replicate all aspects of in vivo liver physiology in terms of immune responses and systemic pharmacokinetics, which play crucial roles in DILI. Consequently, hLiMTs may not capture idiosyncratic or rare hepatotoxic reactions that only occur in the context of a complete organism.

Secondly, while this was the largest benchmarking study, it was still a relatively small number of drugs in each category of DILI risk. Expanding this model to include a broader array of drugs and concentrations would help to validate the findings across more diverse compounds and toxicity profiles.

Another potential limitation is the empirical C2C threshold of 176, which may not extrapolate to other testing environments or drugs with different pharmacokinetics.

Multiparametric assessment of hepatoxicity to enhance predictivity

This step is the first in the journey to revolutionize drug safety screening. Expansion of this work will include validation across a broader range of drugs and refining the model to improve its predictive accuracy for DILI among different classes of drugs. Integrating additional cell types, such as immune cells, could help capture more complex toxicological responses that occur in vivo. Further research may also optimize the IC50ATP-to-Cmax (C2C) threshold to enhance consistency across different drug classes. As these 3D liver microtissue models gain validation, regulatory acceptance and mainstream adoption, they could tremendously reduce the reliance on animal models, facilitating a shift toward more ethical and human-relevant preclinical testing, potentially leading to safer and more effective drug development.

Reference: Fäs L, Chen M, Tong W, et al. Physiological liver microtissue 384-well microplate system for preclinical hepatotoxicity assessment of therapeutic small molecule drugs. Toxicol Sci. Published online October 14, 2024:kfae123. doi:10.1093/toxsci/kfae123

Have you recently published research?

You can raise awareness of your work with an article like this too.

Find out more