One year ago, the U.S. Food and Drug Administration (FDA) announced its intent to phase out animal testing requirements for drug development. Last month, it confirmed it had met its year-one goals, a signal that the shift toward human-relevant models is well underway.
Organoids and other advanced 3D cell models sit at the center of this transition. Their potential is immense: more predictive drug testing, more physiologically relevant disease modeling, and richer datasets for AI-driven discovery. But as the field accelerates, a foundational question emerges: Can the science be trusted?
The missing link in advanced models
Organoids retain the mutations, cellular architecture, and biological behavior of their tissue of origin, qualities that make them far more predictive than traditional 2D cultures or animal models. But that physiological relevance is only as reliable as the materials and methods behind it.
As research innovation accelerates, what lags behind is the infrastructure that makes results trustworthy and transferable across laboratories: authenticated starting materials, harmonized protocols, and integrated data.
Authentication matters more than many researchers realize. Not all cell sources are equivalent. Unverified cell identity, undocumented passage history, or inconsistent donor material introduces variability that no downstream platform can correct.
One analysis found more than 32,000 articles reporting research conducted on misidentified cell lines. This work was cited in an estimated half-million subsequent papers, with HeLa cell contamination among the most well-documented causes.1 If an experiment produces an unexpected result, researchers need confidence that the biology is the variable, not the starting material.
The data challenge compounds this. Organoids generate rich, complex outputs, but those outputs are only meaningful when paired with complete provenance: clinical metadata, genomic annotation, and a documented chain of custody from patient sample to research model. Without it, comparing results across laboratories becomes an exercise in assumption rather than science.
One Nature survey found that more than 70% of researchers had tried and failed to reproduce another scientist’s experiments, with inconsistent materials and incomplete methods reported among the leading causes.2
That risk grows as organoid-derived data increasingly feeds AI-driven analysis and computational modeling. Incomplete or untraced inputs don’t just affect a single experiment. They train and propagate through predictive models, compounding uncertainty at every step.
From sample to science: building a traceable pipeline
Closing the traceability gap requires attention at three distinct levels: the biological inputs, the supporting components, and the data paired to both.
Starting materials
Authentication is the foundation of any trustworthy advanced model. For standard cell lines, this means documented identity through short tandem repeat (STR) profiling, species verification, mycoplasma testing, and sterility checks. These are the baseline quality controls that confirm a material is what it claims to be and behaves as expected. When these reference materials are well-characterized and consistent lot-to-lot, researchers have a reliable platform from which to work. Any unexpected result at this stage points to a protocol issue rather than biological noise.
For patient-derived organoids, the authentication requirements extend further. Identity verification is still essential, but it must be paired with complete clinical metadata, including diagnoses, demographic information, mutation profiles, and treatment history. The Human Cancer Models Initiative, for example, has built a collection of more than 800 models across 28 tumor types, with full clinical and genomic annotation for each model. That level of documentation is what transforms a biological sample into a research-grade tool and makes it possible to interpret results with confidence rather than assumption.
Supporting components
Cells, media, and extracellular matrix all influence how an advanced model behaves, and small inconsistencies in any of them can produce results that are difficult to compare across experiments or platforms. When these components are harmonized (selected and validated to work together across multiple systems), researchers can aggregate and compare data across platforms with confidence.
Organoids present unique challenges and require more experienced handling than cells culture in 2D. Variability in something as fundamental as cell counting can introduce noise that undermines reproducibility from one plate to the next. Developing methods to minimize that variability and delivering materials as close to assay-ready as possible are among the most practical things the field can do to accelerate adoption. ISO-certified bioresource centers provide the quality framework that makes this possible, ensuring that every step of the process, from receipt of material through propagation, quality control, and distribution, is documented, validated, and reproducible across lots.
Paired data
A biological material becomes substantially more valuable when paired with complete provenance, which includes clinical metadata, genomic annotation, and a documented record of how the model was derived, propagated, and quality-controlled. The value compounds further when that data is linked to experimental outcomes. In our work with pancreatic cancer organoid models, for example, pairing genomic profiles with drug sensitivity data revealed distinct response patterns across models carrying different KRAS mutations. These findings would have been difficult to interpret without complete molecular annotation tied to each model.3
As organoid-derived data increasingly feed AI and computational biology pipelines, the integrity of this traceability chain becomes even more consequential. Predictive models are only as reliable as the biological inputs that train and validate them. When the chain of custody from the patient sample to the characterized model to the curated data set is intact, the science that follows can be trusted.
Trust is the foundation
Regulatory momentum is building, the technology is maturing, and the scientific case for human-relevant models is compelling. What determines whether that full potential is realized is the quality of the underlying infrastructure.
As the number of research organizations developing 3D-cell culture and organoid technologies has grown worldwide, fragmentation and inconsistency have emerged as unwelcome byproducts. Protocols vary from lab to lab, reagent choices differ, and culture conditions are rarely standardized across sites. Without a common infrastructure to harmonize these approaches, comparing results across institutions remains difficult, and reproducibility and scientific advancement suffer.
Democratizing organoids means more than expanding access. It means ensuring that a model distributed anywhere in the world performs consistently, that the cells are authenticated, the provenance is documented, and the data is sufficiently complete to be meaningful. It means building a global biorepository infrastructure in which materials and data converge, and where researchers can trust what they receive because every step of the process has been validated and recorded.
The institutions doing this work, those that are establishing qualification frameworks, harmonizing biological inputs, and pairing materials with rich, annotated datasets, are laying the groundwork for a new era of reproducible, human-relevant science where traceability is the foundation on which ground-breaking research is built.
References
1. Horbach SPJM, Halffman W. The ghosts of HeLa: How cell line misidentification contaminates the scientific literature. PLoS ONE. 2017;12(10):e0186281.
2. Baker M. 1,500 scientists lift the lid on reproducibility. Nature. 2016;533:452–454.
3. Friend S., Graziano M., Demarath R., Clinton J. Pancreatic cancer organoids from the Human Cancer Models Initiative biobank reflect disease genotypes, capture patient heterogeneity, and are amenable to therapeutic screening. Poster presented at: American Association for Cancer Research Annual Meeting; 2024; San Diego, CA.
Carolina Lucchesi, Ph.D., is Principal Scientist and Head of Microphysiological Systems at ATCC, where she leads the development and commercialization of the organization’s MPS program. With more than 15 years of experience in organ-on-chip research – including foundational work at the Wyss Institute at Harvard and Emulate, she focuses on accelerating adoption of advanced in vitro models for drug discovery and disease research.