iPSC-Derived Hepatocytes vs Primary Hepatocytes vs HepG2: Which Model for Your Study?

Choosing a hepatocyte model for drug discovery looks straightforward — until you actually make the choice. Three options dominate the field: HepG2 cell lines, primary human hepatocytes, and iPSC-derived hepatocytes. Each has genuine strengths. Each has well-documented failure modes. And for decades, researchers have navigated this decision on astudy-by-study basis, accepting trade-offs as an unavoidable fact of life.

That calculus is changing. The maturity of iPSC-derivedhepatocyte technology has reached a point where the trade-offs are no longer symmetrical. This post lays out an honest comparison of all three models — what each does well, where each falls short, and why pixHep, our iPSC-derived hepatocyte, is increasingly the answer to questions that none of the other systems can fully resolve.

The Three Models at a Glance

HepG2 and Hepatocellular Carcinoma Lines

HepG2 is a human hepatocellular carcinoma cell line derived in 1979 and in continuous laboratory use for over four decades. Its practical appeal is obvious: robust, inexpensive, commercially available, easy to culture, and highly reproducible. It grows indefinitely. It requires no donor tissue.

The biological limitations are equally well-documented.CYP450 activity in HepG2 is roughly 1,000-fold lower than in primary human hepatocytes (Westerink & Schoonen, 2007). HepaRG, a more differentiated carcinoma-derived line, shows better CYP expression but still falls short of primary benchmarks across key enzyme families.

Compounds that are non-toxic but converted to reactive metabolites by CYP enzymes will appear safe in HepG2 screens — and only reveal their toxicity in a more metabolically competent system. Given that a significant fraction of clinically observed DILI is metabolite-mediated, this is not a minor limitation.

HepG2 remains genuinely valuable for high-throughput primary cytotoxicity screening and mechanistic studies of specific pathways where metabolic competence is irrelevant. Beyond those applications, it is a convenient approximation — not a physiologically relevant model.

Primary Human Hepatocytes

Primary human hepatocytes, isolated from donor liver tissue by collagenase perfusion, have been the reference standard for hepatic metabolism studies for decades. At the time of isolation, they express the full complement of Phase I and Phase II metabolic enzymes, functional transport proteins, and bile acid handling machinery at levels comparable to the intact liver. Regulatory agencies have accepted primary hepatocyte data as the basis for IND-stage drug metabolism characterisation for many years.

The limitations are both practical and biological. Donor tissue is scarce, irregular in availability, and variable in quality. Lot-to-lot variability between donors is substantial — not because of poor science, but because donors are genuinely different people with different genetic backgrounds, disease histories, and drug exposures.

The biological problem is dedifferentiation. Within 24–48hours of plating, CYP expression begins to decline. By 72 hours, many Phase I enzymes are substantially reduced. Sandwich culture systems and 3D spheroid configurations slow this process, but do not resolve the underlying biology.

The deeper limitation is fixed: primary hepatocytes cannot be engineered. You cannot introduce disease-specific mutations or generate isogenic pairs. What the donor gave you is what you work with.

pixHep: What iPSC-Derived Hepatocytes Actually Deliver

Human iPSC-derived hepatocytes are generated by directed differentiation through definitive endoderm, hepatic specification, andhepatocyte maturation — recapitulating embryonic liver development in vitro. The resulting cells express albumin, alpha-1-antitrypsin, HNF4A, and other canonical hepatocyte markers, with measurable CYP activity, urea synthesis, and bile transport function.

The fundamental advantage of iPSC-derived hepatocytes is not superiority in any single assay. In most direct comparisons of metabolic enzyme activity, well-differentiated primary hepatocytes outperform iPSC-derived cells — particularly for CYP3A4 and CYP2D6. The fundamental advantage is everything else.

Maturity That Matters: ASGR1 as a Functional Benchmark

A frequent critique of iPSC-derived hepatocytes is insufficient maturity. This critique has merit in aggregate — but it conflates all iPSC-derived hepatocytes as a class, ignoring the extraordinary variation in protocol quality that defines this field.

pixHep is highly characterised and mature, demonstrating functional membrane localisation and activity of the Asialoglycoprotein Receptor 1 (ASGR1) — a benchmark that distinguishes truly mature hepatocytes from partially differentiated intermediates. ASGR1 is aliver-specific receptor responsible for glycoprotein clearance and a well-validated marker of hepatocyte maturity. Its correct membrane localisation and function in pixHep reflects the same receptor profile found in healthy adult human hepatocytes.

This level of maturity matters practically: it enables researchers to study how a compound is received by functionally competent hepatocytes in vitro, with confidence in the biological relevance of the data generated.
‍

The maturity gap between iPSC-derived and primary hepatocytes is real — but it is not fixed. It is a function of differentiation protocol quality, and it is narrowing. Work from our team demonstrated that optimised protocols can generate iPSC-derived hepatocytes with CYP450 expression and activity comparable to primary cells across multiple enzyme families.

Scalability Without Sacrificing Consistency

Once an iPSC line is established and a differentiation protocol is validated, cells can be produced in reproducible quantities from the same genetic source — indefinitely. Lot-to-lot variability is dramatically lower than for primary hepatocytes sourced from different donors, because the genetic source is fixed.

For high-throughput drug discovery, this matters enormously. Studies that require hundreds of thousands of cells across multiple timepoints and conditions can be planned with confidence, rather than constrained by donor availability or batch inconsistency.

Long-Term Culture: The Chronic Toxicity Advantage

Unlike primary hepatocytes, which begin dedifferentiating within hours of plating and are typically unsuitable for studies extending beyond three to five days, pixHep maintains phenotype over extended periods —weeks, not hours.

This enables study designs that primary hepatocytes simply cannot support: chronic toxicity modelling, disease progression studies, repeated compound dosing regimens, and experiments investigating compounds with low clearance rates where extended exposure is necessary to observe meaningful pharmacological or toxicological effects.

Genetic Control: The Capability Primary Cells Cannot Match

iPSC-derived hepatocytes can be produced from any donor and engineered with precision. CRISPR editing enables the introduction of specific disease-causing mutations into otherwise isogenic backgrounds, generating paired wild-type and disease variant lines. Patient-derived lines capture the full polygenic complexity of disease-associated backgrounds.

This genetic controllability underpins the most important applications that primary hepatocytes cannot serve: mechanistic studies of inherited metabolic liver diseases, pharmacogenomic studies of drug response variation, and validation of genotype-specific drug targets.

‍Enabling Predictive Models: Why Consistency Is the Foundation

Predictive ADMET and toxicity models are only as reliable as the data they are trained on. When that data is generated across multiple donor lots — each with its own genetic background, enzymatic activity profile, and culture behaviour — the biological noise compounds. Models trained on primary hepatocyte data carry that donor variability embedded in their predictions, making it difficult to distinguish a true compound signal from a donor effect.

iPSC-derived hepatocytes change that equation. Because pixHep is produced from a fixed genetic source with a validated differentiation protocol, the biological inputs to any predictive model are consistent across batches and timepoints. That consistency means that variability in the data reflects the compound — not the cell source. It is the foundation on which meaningful structure-activity relationships, toxicity classifiers, and multi-timepoint mechanistic models can actually be built.

As the field moves toward AI-assisted drug discovery and high-content phenomics, the value of a reproducible, scalable cell source becomes structural rather than merely convenient.

Direct Comparison

The table below summarises the key attributes of each model. The pixHep column reflects our current generation of iPSC-derived hepatocytes.

Which Model for Which Question

Use HepG2 or HepaRG when:

•      High-throughput primary cytotoxicity screening across large compound libraries is the priority
•      Cost and throughput are the primary constraints
•      Compounds are not expected to require metabolic activation
•      The goal is rapid triagebefore more resource-intensive assays

 Use primary hepatocytes when:

•      Regulatory acceptance of drug metabolism data is required
•      CYP inhibition and induction profiling or reaction phenotyping is the specific objective
•      The study duration is compatible with the 48–72 hour dedifferentiation window
•      The full metabolic complexity of a freshly isolated adult hepatocyte is specifically required

Use pixHep when:

•      You need to model aspecific disease genotype or patient population
•      Long-term culture is required — chronic toxicity, disease progression, or repeated dosing
•      Multi-donor coverage isneeded to capture population-level variability in drug response
•      CRISPR-engineered isogeniccontrols are required for mechanistic studies
•      You want reproducibility atscale without donor-dependent variability
•      You are combininghepatocyte biology with high-content imaging or Cell Painting phenomics

Increasingly, the most informative experimental designs combine all three. HepG2 screens narrow the compound space. Primary hepatocyte assays characterise metabolic liability. pixHep — in multi-donor anddisease-specific configurations — provides the population-level andgenotype-specific data that neither of the other systems can generate.

The Direction of Travel

The trajectory of the field is clear. Each successive generation of iPSC-derived hepatocyte differentiation protocols has moved the maturity profile closer to primary tissue. The biological gap that was once the central objection to iPSC-derived hepatocytes is narrowing with every iteration.

At the same time, the advantages that iPSC-derived hepatocytes offer — and that primary hepatocytes structurally cannot match —are not temporary. Genetic control, indefinite scalability, multi-donor capability, long-term culture stability, and compatibility with morphological profiling are features of the technology. They do not erode with time.

The field is moving toward iPSC-derived hepatocytes not because they are already superior to primary cells in every assay, but because the trajectory of improvement points clearly in that direction — and because the things they can do that primary hepatocytes cannot are increasingly the things that drug discovery programmes most need.

pixHep is our contribution to that trajectory: highly characterised, functionally mature, scalable, and designed for the complexity of modern drug discovery.

Want to discuss which hepatocyte model is right for your study?

Speak with the pixlbio scientific team — visit Contact Us or explore our pixCells and disease modelling pages for more information on pixHep.

Tags: iPSC Hepatocytes  · Primary Hepatocytes  ·  HepG2 ·  HepaRG  ·  Drug Discovery  ·  DILI ·  Hepatotoxicity  · CYP450  ·  pixHep