Published in BMC Methods, the study presents a stepwise methodology designed to address the critical need for systematic frameworks in cell selection. As the array of primary and stem cell-derived sources expands, the performance and translational relevance of liver-on-a-chip models depend heavily on the quality and suitability of the cells used to populate them. The team aimed to create a reproducible process that ensures consistent evaluation across different cell lots, thereby improving the accuracy of disease modeling and drug testing in these advanced in vitro platforms.
The study focuses on the liver acinus microphysiological system, a device that replicates key aspects of liver structure and function. This platform integrates microfluidics, three-dimensional cell cultures, and biomimetic environments to create physiologically relevant models for drug safety and disease progression studies. The research team, led by Jacquelyn A. Brown and colleagues, applied this protocol to four distinct lots of primary human hepatocytes. The overarching goals were to develop a pipeline for evaluating cell suitability, compare performance across lots under physiologically relevant conditions, and identify the most informative functional metrics aligned with metabolic dysfunction-associated steatotic liver disease modeling.
The qualification process begins with rigorous donor screening and initial cell viability assessments. The researchers excluded pediatric donors and individuals with known liver disease to preserve the model’s capability to induce disease phenotypes under controlled experimental conditions. Initial testing involved trypan blue viability assays and evaluations of cell attachment and morphology. Hepatocyte lots were required to demonstrate post-thaw viability greater than 80 percent and exhibit cell attachment and spreading within eight hours. Lots meeting these thresholds were considered replateable and advanced to the next phase, where the full biomimetic model was established with the addition of non-parenchymal cell types.
Phase two of the protocol focuses on viability and stability metrics over an eight-day culture period. The team monitored lactate dehydrogenase release to assess temporal cell health, observing an initial spike as the model stabilized, typically by day four, followed by consistent low-level release. At the endpoint, cells were stained with Calcein-AM to provide a direct measure of live, metabolically active cells. The study defined a threshold of at least 80 percent Calcein-AM positivity as acceptable, noting that lower levels correlated with reduced albumin production and diminished cell-to-cell border interactions. One lot was ultimately rejected during this phase due to failing the Calcein-AM metric under late-stage metabolic syndrome conditions, highlighting the importance of multi-parametric acceptance criteria.
Phase three evaluates the performance of hepatic functional assays, covering protein synthesis, ammonia detoxification, and metabolic enzyme induction. The researchers measured urea production, which serves as a marker for mitochondrial activity and ammonia detoxification. The acceptance threshold for urea production was set at 20,000 nanograms per day per million hepatocytes between days two and eight under normal fasting and early metabolic syndrome conditions. Additionally, the team assessed cytochrome P450 3A4 activity following treatment with rifampicin, a critical enzyme for drug metabolism. A minimum three-fold induction post-treatment was required for acceptance, with results showing lot-specific variability in response to different media conditions.
Bile canalicular activity, a marker of hepatocellular polarization and transport processes, was also assessed during the functional phase. Using CellTracker Green CMFDA staining under a collagen overlay, the researchers quantified canalicular flow on days five or six. Acceptable performance was defined as quantification of at least two percent image area with canalicular activity. Albumin production, considered the most sensitive indicator of hepatotoxicity and disease progression, was measured across all lots. The data revealed that early metabolic syndrome media consistently supported the highest albumin output compared to normal fasting and late-stage conditions, with peak production occurring between days five and seven.
Phase four involves the induction of metabolic dysfunction-associated steatotic liver disease phenotypes to test the model’s responsiveness. The team assessed steatosis and fibrosis, two hallmarks of disease progression, under normal fasting, early metabolic syndrome, and late-stage metabolic syndrome conditions. Steatosis was quantified by lipid accumulation using LipidTOX staining. The results showed that early and late-stage metabolic syndrome treatments resulted in significant increases in lipid content, meeting the acceptance criteria for three of the four tested lots. One lot failed to exhibit significant lipid accumulation under late-stage conditions, demonstrating the variability in disease-relevant phenotypes across different cell sources.
Fibrosis was assessed by measuring pro-collagen Iα1 secretion into the media and immunofluorescence staining for alpha-smooth muscle actin to evaluate stellate cell activation. The study found that pro-collagen Iα1 levels provided a clearer distinction in fibrotic response compared to alpha-smooth muscle actin, which proved to be a less reliable marker heavily influenced by the specific hepatocyte lot. While some lots showed increased alpha-smooth muscle actin staining, only one lot reached statistical significance under early metabolic syndrome conditions. This variability underscores the complexity of modeling fibrosis and the need for careful interpretation of fibrotic markers in microphysiological systems.
The final phase of the roadmap focuses on reproducibility testing to ensure that the selected cell lots yield consistent results across independent runs. The top two candidates from initial testing were re-evaluated to confirm their stability and reliability. Both lots demonstrated acceptable reproducibility, with one showing slightly superior consistency in most metrics. The intraclass correlation coefficient was used to assess both intra- and inter-study reproducibility, with values classified as excellent, acceptable, or poor based on established thresholds. This step is crucial for advancing the technology beyond academic research into clinical and pharmaceutical applications, where consistent and meaningful biological insights are essential.
The study highlights the limitations of relying solely on rigid performance metrics, which may overlook intrinsic biological variation due to donor genetics or disease state. The researchers propose a fit-for-purpose evaluation framework that balances functional relevance with biological heterogeneity. This approach allows for the systematic identification of how intrinsic donor heterogeneity impacts model outcomes, providing critical meta-information for advancing precision medicine. By generating rigorous, quantitative performance metrics from primary hepatocytes, the team creates a valuable reference for assessing the maturation and disease modeling capabilities of induced pluripotent stem cell-derived hepatocytes.
The development of this standardized protocol addresses a significant gap in the current state of microphysiological systems, which lacks rigorous criteria for cell selection. This limitation has hindered inter-laboratory reproducibility and broader adoption of these platforms. By establishing qualification standards for the cells used to construct the models, the study aims to enhance the physiological relevance and reliability of liver-on-a-chip technologies. The protocol is adaptable to other liver diseases or applications, offering a versatile tool for the research community.
As the field of microphysiological systems continues to evolve, fostering cross-platform consistency while embracing biologically relevant variability will be essential. The implementation of a structured, reproducible roadmap for cell qualification is critical for advancing the field toward more predictive and clinically meaningful models. This work provides a foundation for collaborative science, increasing confidence in the biological conclusions drawn from microphysiological system-based studies and supporting the transition from animal models to human-based in vitro systems.
The liver acinus microphysiological system used in this study incorporates four distinct cell types to mimic the liver acinus structure. Primary cryopreserved human hepatocytes were seeded at densities of 2.70 to 2.85 million cells per milliliter, representing 56 percent of the total cell composition. Liver sinusoidal endothelial cells were included at 1.5 million cells per milliliter, accounting for 22 percent of the model. Kupffer-like cells, derived from the THP-1 monoblast line and treated with phorbol 12-myristate 13-acetate, comprised 18 percent of the population. Human stellate cells, specifically the LX-2 line, made up the remaining 4 percent. This specific cellular ratio aligns with previously published scaling methods designed to replicate physiological liver architecture. The devices were coated with bovine fibronectin and rat-tail collagen I to support cell adhesion and tissue formation. Perfusion was maintained at a flow rate of 5 microliters per hour for eight days, a condition chosen to mimic zone-3 oxygen tension within the liver. This controlled environment allows researchers to observe how different cell types interact under sustained metabolic stress, providing a dynamic platform for evaluating cellular responses over time.
Media formulations were designed to model the progression of metabolic dysfunction-associated steatotic liver disease from a normal fasting state to early and late stages of metabolic syndrome. The normal fasting media served as a baseline for healthy liver function, while the early metabolic syndrome media introduced conditions associated with the initial stages of disease. The late metabolic syndrome media represented a more advanced stage of disease progression, incorporating pro-fibrotic drivers such as transforming growth factor beta-1 and lipopolysaccharide. These media types were applied to the microphysiological systems for the duration of the eight-day experiment. The researchers noted that exposure to these diseased media conditions is sufficient to establish key features of the disease at the appropriate stage. This approach allows for the controlled induction of disease phenotypes in cells sourced from phenotypically healthy donors, ensuring that observed changes are attributable to the experimental conditions rather than pre-existing donor pathology. The distinction between these media types is crucial for understanding how liver cells respond to escalating metabolic stress.
Statistical analysis played a central role in determining the reliability of the observed metrics. One-way analysis of variance was used to compare three or more studies conducted under identical conditions with the same cell lot, while a pooled t-test was applied for duplicate identical studies. The intraclass correlation coefficient was employed to assess both intra- and inter-study reproducibility of the metrics measured at multiple time points. This coefficient evaluates both the magnitude and trend of the signals over time, providing a comprehensive view of model stability. Reproducibility was classified as excellent when the coefficient was greater than or equal to 0.8, acceptable between 0.2 and 0.8, and poor below 0.2. Inter-study reproducibility analysis was conducted following the Pittsburgh Reproducibility Protocol, a standardized framework for evaluating consistency across different experimental runs. This rigorous statistical approach ensures that the differences observed between cell lots are not merely due to random variation but reflect genuine biological variability. By applying these statistical standards, the researchers were able to identify which lots demonstrated consistent performance and which exhibited erratic behavior, thereby informing the final selection process for the microphysiological system.
Subject of Research: Road map for primary hepatocyte qualification in human liver organ models
Article Title: Road map for primary hepatocyte qualification in human liver organ models
Article References: Varmazyad, M., Gavlock, D. C., Castiglione, M. W., DeBiasio, R., LaRocca, G., Reese, C., Vernetti, L. A., Schurdak, M., Stern, A. M., Taylor, D. L., Miedel, M. T., & Brown, J. A. (2026). Road map for primary hepatocyte qualification in human liver organ models. BMC Methods, 3(1), Article 6. https://doi.org/10.1186/s44330-026-00058-7
Image Credits: AI Generated
DOI: 10.1186/s44330-026-00058-7
Keywords: Road, primary, hepatocyte, qualification, human, liver, organ, models, scientific research
News Source: Gregory Coleman. (November 13, 2026). New Protocol Standardizes Hepatocyte Selection for Liver Chips. Scienmag.
