NAMs for DILI prediction: What’s driving the change?

Drug-induced liver injury (DILI) remains a leading cause of late-stage drug failure, making reliable in vitro hepatotoxicity testing a priority in preclinical toxicology. Yet, the cells powering these assays carry fundamental constraints.

Immortalised hepatocyte cell lines, such as HepaRG, have long offered a reproducible, off-the-shelf option for early screening. However, their limited CYP activity and single-donor origin limit their relevance for DILI prediction, which is why the field commonly uses primary human hepatocytes (PHH).

PHH are widely regarded as the gold standard for in vitro liver modelling, offering broader CYP activity and metabolic diversity than alternative cell sources. In practice, though, they present considerable challenges. In traditional 2D culture, PHH de-differentiate within 3-4 days, losing metabolic function. This rules out repeat-dose and chronic studies, pushing researchers towards 3D systems to sustain functionality long enough to run meaningful assays.

Further, donor-to-donor variability adds another layer of complexity. PHH are derived from donated liver tissue, and the quality of that starting material can vary widely. For instance, the tissue may be steatotic, ischaemic, or compromised in ways that diminish metabolic functionality from the outset. Consequently, each lot must be validated before use, draining time and budget. When the lot runs out, the process starts over. 

While pooled donor approaches can broaden sample diversity for larger studies, they also mask individual metabolic profiles, making it harder to interpret assay results with confidence.

iPSC-derived hepatocytes offer a scalable, donor-independent alternative to PHH. Existing models, however, are considered functionally immature, exhibiting insufficient CYP activity and persistent AFP expression. As a result, adoption has remained limited among toxicologists accustomed to PHH performance benchmarks.

Meanwhile, the regulatory push towards non-animal safety testing is gaining pace. The FDA Modernization Act 2.0¹, the gradual adoption of the 3Rs framework (Replace, Reduce, Refine)² and the UK government's strategy³ on the replacement of animals in science are accelerating demand for human-relevant in vitro models.

The collective result is a rapidly evolving landscape for New Approach Methodologies (NAM) in DILI prediction, and a sharper focus on cell source quality, both of which we explore in this article.

Current NAM platforms for in vitro DILI assessment

From organ-on-a-chip to spheroid co-cultures, preclinical toxicology platforms for DILI prediction have advanced considerably. Here’s a closer look at the key players, what they’re doing well, and where shared challenges remain.

1. Emulate’s Liver-Chip 

Using a microfluidic organ-on-a-chip device, the Liver-Chip recreates the human liver sinusoid by combining PHH with liver sinusoidal endothelial cells, Kupffer cells, and stellate cells under continuous flow.

Strengths

• The multi-cellular environment supports advanced mechanistic insights

• Demonstrated capacity for 7-day repeat-dose studies

• The ‘Organ Chips in Space’ initiative has drawn high-profile attention to the field

• A landmark study⁴ reported 87% sensitivity for identifying hepatotoxins that had passed animal testing

Limitations

• Relies on PHH, which remain subject to donor variability and batch revalidation requirements

• Study duration is extended to 14 days in vitro but is ultimately constrained by PHH de-differentiation, even within the chip environment

• The chip-based format limits throughput, making the platform more suitable for mechanistic investigation rather than large-scale screening

2. CN Bio’s Liver-on-a-Chip

Combining PHH and Kupffer cells in a perfused liver-on-a-chip format, the CN Bio system models hepatic function under continuous media flow.

Strengths

• Perfusion-based culture promotes high metabolic activity and extended culture longevity

• Provides immune-mediated DILI capability, a key differentiator for studying liver injury

• Cross-species models (rat, dog) for translational studies

Limitations

• Relies on PHH, with CN Bio noting that pre-mixed pooled hepatocytes can generate poor quality microtissues in organ-on-a-chip systems⁵

• Recommends testing individual donor lots rather than pools, adding validation time and cost

• Donor-matched non-parenchymal cells are needed, further constraining supply chain flexibility

3. inSphero’s 3D liver spheroids

Taking a spheroid-based approach, inSphero’s 3D model uses pre-aggregated co-culture microtissues combining PHH from a 10-donor pool (5 male, 5 female) with non-parenchymal cells in assay-ready 96- or 384-well plates. 

Strengths

• Assay-ready format reduces hands-on setup time; spheroids arrive pre-formed and quality-controlled

• 10-donor pool captures broader metabolic diversity than single-donor approaches

• Up to 4 weeks of viability; supports repeat-dose and chronic study designs

Limitations

• Despite standardised production, underlying donor variability and cell quality persists, and each PHH lot still needs to be validated before use

• When a validated PHH lot is exhausted, revalidation of a new lot is required – a transition that can introduce variability to the user’s results, even with rigorous QC in place

• The 3D spheroid format, while extending PHH viability, adds model complexity compared with standard 2D workflows

4. Mimetas’ high-throughput microfluidic liver model

Using gravity-driven perfusion across a microfluidic chip, the Mimetas platform co-cultures iPSC-derived hepatocytes with endothelial and Kupffer cells.

Strengths

• One of the few platforms to adopt iPSC-derived hepatocytes for DILI screening at scale
• Stable 15-day viability with albumin and urea secretion, validated against compound libraries⁶
• 96-well microfluidic format supports high-throughput and automated workflows
• Multi-organ capability (liver, lung, brain) on a single-plate format

Limitations

• iPSC-derived hepatocytes carry the industry-wide concern around functional maturity, a barrier to broader adoption by toxicologists accustomed to PHH performance benchmarks

Hepatic cell source: A shared constraint across current DILI prediction models

Across existing platforms, enormous engineering effort goes into compensating for the limitations of the cells going into them. As a result, the field faces a set of trade-offs:

• Immortalised cells, such as HepaRG, provide off-the-shelf availability and ease of use, at the cost of narrow metabolic representation and incomplete CYP coverage
• PHH deliver mature hepatic function at the cost of donor variability, finite supply, and rapid de-differentiation in 2D
• Traditional iPSC-derived hepatocytes offer donor independence and scalability at the cost of functional maturity

In either case, 3D architecture is typically needed to sustain repeat-dose or chronic studies, adding complexity and limiting throughput. Toxicologists now recognise that platform innovation alone cannot outpace cell source limitations.

The unmet need is clear – human hepatocytes that are healthy, mature, characterised, lot-consistent, and stable in 2D for the duration of a long-term study, without forcing researchers into a 3D workaround.

Consistent, scalable, and functionally mature: ioHepatocytes for DILI prediction

Rather than building a new platform, bit.bio has taken a different approach: focusing on the cell itself.

Our ioHepatocytes are developed using opti-ox™, a proprietary deterministic cell programming technology that generates consistent, defined human iPSC-derived hepatocytes. The outcome is a cell that combines the functional maturity toxicologists expect from PHH with the scalability and consistency that have, until now, been out of reach.

How ioHepatocytes address current constraints of in vitro liver models

• Functional maturity: ioHepatocytes display CYP activity comparable to PHH. Their AFP expression decreases over time in culture, consistent with a maturing, adult-like hepatocyte phenotype.

• Lot-to-lot consistency: Because opti-ox programming is deterministic, each manufacturing run produces cells with the same well-characterised profile, eliminating donor variability and batch revalidation associated with PHH.

• Extended 2D longevity: ioHepatocytes maintain functional stability for over two weeks in standard 2D culture, enabling repeat-dose study designs without the need for 3D scaffolding or perfusion systems.

• Scalability: Manufactured at scale, bit.bio’s proprietary opti-ox™ technology can deliver billions of ioHepatocytes per production run, supporting high-throughput screening in standard plate formats.

“These characteristics position ioHepatocytes as a complementary cell source that could enhance existing organ-on-a-chip and spheroid platforms while supporting scalable, higher-throughput 2D screening. They also give preclinical toxicologists the opportunity to conduct longer-term studies without reverting to 3D formats to maintain cell viability.”


Tom Harris-Brown
Senior Portfolio and Partnerships Manager, Toxicology

ioHepatocytes demonstrate sustained albumin secretion in 2D culture. (A) ELISA quantification of albumin secretion confirms a time-dependent increase in metabolic competency during cell maturation. Albumin production reaches >10µg/million cells/24hr by day 11 post-thaw, and doubles to >20µg/million cells/24hr by day 17, establishing a robust and healthy primary hepatocyte-like functional profile capable of maintaining metabolic competency over time. (B) Immunofluorescence at day 11 and 17 confirms robust albumin expression. 

ioHepatocytes detect DILI-relevant cytotoxicity in 2D culture. ioHepatocytes and primary human hepatocytes (PHH) were exposed to 7 distinct test compounds across a broad concentration gradient for 24 hours. Cell viability was determined relative to vehicle control. Upon exposure, ioHepatocytes exhibit a clear, dose-dependent cytotoxicity profile that correlates strongly with PHH benchmark performance. Both models successfully stratify the compounds based on clinical DILI severity - capturing distinct cytotoxic drops for severe/moderate toxins (e.g., Sunitinib, Amiodarone, Ketoconazole) while displaying a stable viability profile for Valproic acid, used as a control compound. Data generated from cells in continuous culture (non-cryopreserved).

How ioHepatocytes compare across key cell source attributes

What this means for DILI screening workflows

The NAM ecosystem for preclinical toxicology has matured significantly, and regulatory frameworks are beginning to catch up. Organ-on-a-chip devices, spheroid platforms, and microfluidic systems have each advanced our capacity to model hepatotoxicity in vitro. However, platform innovation alone can only go so far. The next significant leap in DILI predictivity will come from closing the gap between platform capability and cell quality. 

As more predictive, human-relevant in vitro tools, such as ioHepatocytes, enter preclinical toxicology workflows, the promise extends beyond better data. It is a step towards advancing the 3Rs and moving safety toxicology closer to a non-animal paradigm – one built on human cells engineered to be consistent, scalable, and fit for purpose.

Learn more about ioHepatocytes and how they support hepatotoxicity assays.

References

1. S.5002 - FDA Modernization Act 2.0 (Congress.gov)
2. Regulatory Applications of 3Rs, National Toxicology Program
3. Technology alternatives to animals in life sciences research, UK Parliament
4. Ewart, L., Apostolou, A., Briggs, S.A. et al. Performance assessment and economic analysis of a human Liver-Chip for predictive toxicology. Commun Med (2022)
5. A guide to pre-validating primary cells for use in Organ-on-a-chip assays, CN Bio
6. Bircsak KM, et al. A 3D microfluidic liver model for high throughput compound toxicity screening in the OrganoPlate®. Toxicology. (2021)