Cross-Species Liver Study Finds Limited Added DILI Sensitivity from Perfusion
Primary hepatocytes from humans, monkeys, rats and dogs were compared in monolayers, static aggregates and perfused microtissues. Static human cultures detected fialuridine responses, while the low-throughput perfused comparison showed weaker, nonsignificant biochemical changes.
Separating Species Choice from Culture Complexity
A Toxicological Sciences study led by Texas A&M researchers compares primary human, cynomolgus-monkey, Wistar-rat and beagle-dog hepatocytes in three culture formats. The question is whether static monolayers, static spheroid-format cultures or perfused spheroids better maintain liver function and reveal species-dependent drug-induced liver injury, or DILI.
The comparison ran for fourteen days. Drug exposure began on day six, with repeated dosing. Chlorpromazine, bosentan and fialuridine were tested in the static formats; the perfused HUMIMIC Chip2 experiment tested only 100 μM fialuridine because of limited throughput. Thus, the three-compound static comparison and the single-compound perfused comparison have different pharmacological coverage.
Cell Source Affected Tissue Formation and Function
Human hepatocytes formed compact spheroids, whereas monkey, rat and dog cells formed looser aggregates in static microwells. Perfusion improved organization in some animal preparations without producing equivalent compact spheroids across species. The human lot had been selected for spheroid-forming ability; the animal lots had not undergone the same qualification.
Human albumin secretion was relatively robust across formats. Perfusion gave modest functional improvements, with levels broadly comparable to static spheroids. Animal hepatocyte function generally declined with time, and transcriptional changes associated with transfer into culture were larger than many later platform differences. By day fourteen, profiles within a species had largely converged across formats.
Many absorption, distribution, metabolism and excretion-related transcripts declined during culture, particularly in rat and dog cells. Retained albumin or urea production therefore did not establish that every metabolic pathway remained at its original activity. The study measured expression patterns but did not directly test the functional consequence of every transcript change.
Static Cultures Captured Distinct Drug Profiles
Chlorpromazine produced the most consistent injury and functional changes in human hepatocytes, with more variable responses in the animal preparations. In static spheroids, falling injury-enzyme signals in some rat and dog conditions could reflect loss of viable cells. A lower released-enzyme value was therefore not necessarily evidence of less injury.
Bosentan produced pronounced responses in human and monkey cells, while rat and dog responses were more variable. Human and monkey monolayers showed reduced albumin and increased lactate dehydrogenase; aspartate aminotransferase increased clearly in the human comparison. The study interprets these differences alongside the compound’s known species-dependent liver liabilities.
Fialuridine produced time- and concentration-dependent injury-marker increases and reduced albumin in human monolayers. In static human spheroids, albumin declined without significant increases in the two injury enzymes. Animal-cell responses were less consistent. Static conditions used three replicates per condition, so these are within-preparation comparisons rather than broad estimates of donor-population performance.
Perfusion Did Not Produce a Stronger Fialuridine Signal
The perfused experiment used two replicates per condition. Fialuridine-associated biochemical trends did not reach statistical significance, and chip-to-chip variability was high. The result offers limited support for added sensitivity from perfusion under the tested conditions, despite maintenance of baseline function.
Transcriptomics added mechanistic resolution. At 100 μM fialuridine, human monolayers and static spheroids showed 1,705 and 1,724 differentially expressed genes, respectively, using the study’s specified fold-change and false-discovery criteria. About two-thirds overlapped between those static human formats. Responses were attenuated in perfused human spheroids, with fewer significantly enriched pathways.
The clearest human pathway changes concerned xenobiotic handling, lipid and bile-acid metabolism and inflammatory programs. Sample loss prevented some rat static-culture transcriptomic comparisons. Explanations for muted perfused responses include exposure kinetics, possible protective effects of flow and limited statistical power; the experiment does not distinguish them.
A Lot-Specific Comparison with Practical Consequences
The study used a single donor for the human, monkey and dog preparations and a pooled rat lot. Its unusually functional human monolayer preparation and spheroid-qualified human cells may influence the platform comparison.
The authors favor matching complexity to the question: scalable static assays for concentration–response testing, with perfused systems reserved for mechanisms where flow or multicellular organization supplies a demonstrated advantage. Broader donor and compound panels, together with measured exposure, are needed to test whether that advantage persists beyond the present preparations.