Following the passing of the FDA’s Modernisation Act 2.0 in 2022, and subsequent announcement in April 2025 to phase out animal testing requirements for monoclonal antibodies (and other drugs), in favour of more human-relevant methods – the use of New Approach Methodologies (NAMs), including Organ-on-a-chip (OOC), is now encouraged and incentivized by the world’s largest regulator for evaluating drug safety and effectiveness.

Despite this progress, replacing animals in safety testing is challenging, since animal models provide many organ-specific readouts that are not fully reproductible in vitro, or modeled in silico. However, there are clear contexts of use where OOC (also known as microphysiological system) data can be used to supplement animal data and improve safety risk assessments today, as we begin the process of switching to a NAMS-focused drug development.

Most medications are metabolised in the liver, which makes it susceptible to adverse drug reactions. A publication by Singh et al., 2025  highlights that 55 medications were approved by the US FDA in 2023, with 22 (40%) having evidence of liver toxicity in the drug label or clinical trial results. So, for some, why was Drug-induced liver injury (DILI) only identified in human studies?

Two retrospective reviews found the positive concordance between liver toxicity in animal tests and clinical trials was 55% (Olson et al., 2000); and 33% (rats), 27% (dogs), and 50% (monkeys) (Monticello et al., 2017).  The main reason relates to interspecies differences in drug disposition, however, there are other contributing factors, as reviewed by Taylor et al., 2025.

This problem is further exacerbated by newer human-specific drug modalities for which animal testing is less suited. These reasons necessitate the need for pre-clinical toolbox modernisation with New Alternative Modalities (NAMs), including OOC, to deliver improved in vitro to in vivo translatability (IVIVT).

How OOCs improve in vitro to in vivo translatability and demonstrate their capability

Human liver-on-a-chip models are utilized to predict and gain a mechanistic understanding of DILI. Their enhanced human relevance provides a means to overcome the limitations of animal models, which can miss some instances of idiosyncratic DILI, plus they provide a viable path forward for new drug modalities.

One approach used by OOC vendors to prove their enhanced translatability is by studying flawed drugs that made it to market. However, unless unsuited, preclinical testing involves testing in animal models. Data discrepancies between human in vitro and animal in vivo studies make the final assessment of drug safety in humans challenging. So, do we need to take a step back and more effectively translate data between in vitro animals and in vivo animal studies to address this gap?

Regulatory guidelines

In preclinical testing, the drug developer is responsible for the choice of in vivo species. Factors, including the drug target and how closely the animal species mimics human responses, influence decision-making, as does previous experience from past trials.

Although times are changing, international guidelines currently specify the use of at least one rodent and one non-rodent species. Rats and dogs are most frequently used in toxicology testing, particularly for small molecules, with non-human primates (NHPs) being common for newer drug modalities. All pre-clinical data needs to be reported in regulatory filings.

Use of cross-species OOC for enhanced in vitro to in vivo translatability

For this reason, CN Bio has been developing preclinical rat and dog Liver-on-a-chip models as translational tools. These tools enable comparative studies to flag interspecies differences early, better inform in vivo study design, and support go/no decision-making. Insights from these advanced in vitro models help to prevent drugs that are potentially safe in humans from being dropped unnecessarily from the pipeline and de-risk those that aren’t adequately flagged by animals from progressing into preclinical testing. 

Sitaxentan, a pulmonary hypertension medication, was removed from the market in 2010 owing to concerns over liver toxicity. A review by Owen et al., 2012 examined preclinical animal species studies, where the severity of the toxicity was missed. The authors highlighted that it was not a deficiency in the preclinical package that resulted in the missed toxicity and translational errors; rather, it was a shortcoming of the preclinical species to predict human outcomes or the mechanistic details of toxicity.

To demonstrate the translatability of cross-species DILI assays, we have been dosing human and cross-species models, cultured using the PhysioMimix^® Core System and Multi-chip Liver-12 plates, with drugs that have known interspecies differences, including sitaxentan. Leveraging clinical markers, including alanine aminotransferase/ aspartate aminotransferase (ALT/AST), this method provides the ability to rank order drugs by safety risk across commonly used species before the preclinical phase.

Additionally, a collaborative team of academic and industrial researchers has published their work developing and evaluating human, cynomolgus monkey, dog, and rat Liver-on-a-chip models for DILI studies using PhysioMimix (Naga et al., 2025).

Their impactful publication was picked up by the editorial team at Science Where Do We Stand With “Liver-on-a-Chip” Technology?, who cite “This new paper is one of the best I’ve seen in this area” with the author stating “The company behind these is one of the major suppliers of advanced assay systems of this sort, and to the best of my knowledge, this is pretty much the state of the art for what you can buy to run your own assays with”.

Led by Texas A&M University, the publication compares Liver-on-a-chip data generated using Multi-chip Liver 12 plates to “classic” 2D cell culture assays (in 96-well plates) to compare the accuracy of PhysioMimix microphysiological systems in detecting liver toxicity.

Following exposure to a variety of compounds, known to cause DILI, including bosentan, fialuridine, and chlorpromazine that exhibit inter-species differences, the results of their study cite that CN Bio’s PhysioMimix System “represents a sensible model for longer-term in vitro studies, particularly in cases where hepatotoxicity may arise through complex or delayed mechanisms”. It recommends incorporating transcriptomics profiling into analysis workflows, rather than relying on liver functionality and health biomarkers (albumin, urea, bile acids, AST, ALT, etc.), for which PhysioMimix is particularly well-suited, with large amounts of recoverable microtissue.

The authors cite the cost and throughput of running these types of experiments as limitations; however, CN Bio has since developed a higher-throughput Liver-48 plate to reduce Liver-on-a-chip costs by 75% and increase throughput up to 288 samples per run.

Cross-species OOC models align with 3Rs objectives

The approach also supports 3Rs objectives and ethical considerations, including the length of testing required, cost, and availability of models (especially for NHPs). It facilitates the responsible use of animals, as hundreds to thousands of in vitro tests can be performed per donor, and their insights safeguard future animal use.

CN Bio continues to qualify these models to understand their in vitro to in vivo translatability potential. Watch our on-demand webinar to learn more, subscribe to our newsletter via our website or follow us on LinkedIn for updates as they happen.

For immediate access to our cross-species (human, rat, dog) DILI assays, contact the CN Bio Contract Research Services team. Alternatively, visit our PhysioMimix DILI assay webpage to discover how to predict human and animal liver responses to acute or repeated drug dosing in your laboratory.

This article was originally published in Scientist Live on 11 April 2025 and has since been modified to reflect further regulatory announcements by the FDA on April 10th (2025), and to reference more recently published data.