Organoids and Lab-Grown Models: Could They Improve How We Test Treatments?

Picture a speck of jelly in a plastic dish. Inside it, a single stem cell taken from a mouse intestine divides, and divides again, and over days folds itself into a hollow bud with pockets and finger-like projections, a rough miniature of the gut lining it came from. Nobody sculpted it. In 2009 a team led by Hans Clevers reported exactly this in Nature: intestinal “crypt-villus” units could build themselves from one stem cell, with no supporting tissue around them.

That speck was an early organoid, and the idea it demonstrated (that cells carry enough instructions to organize themselves into tissue-like structures) has since become one of the busiest corners of biomedical research. The question this article asks is a practical one: could organoids and other lab-grown human models help us test treatments better than the tools we rely on today?

Plain-English: An organoid is a tiny, three-dimensional cluster of human or animal cells, grown from stem cells, that self-organizes to mimic some of the structure and function of a real organ. It is a model of an organ, not an organ.

What organoids are, and what they are not

Organoids usually start from one of two sources. The first is adult stem cells, the repair cells that many tissues keep in reserve, as in the 2009 gut experiment. The second is pluripotent stem cells, cells that can in principle become almost any cell type, including stem cells made by reprogramming ordinary skin or blood cells.

In 2013 a team led by Madeline Lancaster and Juergen Knoblich used pluripotent stem cells to grow what they called cerebral organoids, millimeter-scale structures that formed distinct brain-like regions. Using cells from a patient with microcephaly (a condition in which the brain is abnormally small), they saw neurons forming too early, a clue to the disease that had been hard to study in mice.

Those results are real, and so are the limits. A brain organoid is not a brain: it has no senses, no blood supply and nothing like the organization of a living nervous system. The same goes for organoid “livers”, “kidneys” and “lungs”. Each captures some features of the tissue, often at a developmental stage closer to fetal than adult.

Why drug testing needs better models

Before a new medicine is given to people, it passes through what the FDA calls preclinical research: laboratory and animal testing designed to answer basic questions about safety. Animal studies remain valuable, but mice, rats and even monkeys are not small humans. Their metabolism, immune systems and organ biology differ in ways that can hide a danger or exaggerate a benefit.

Flat layers of human cells in a dish, the traditional alternative, miss the three-dimensional architecture and cell-to-cell conversations of real tissue. Organoids, and their engineered cousins, aim to sit in between: human biology, with some tissue-like complexity, at a scale a laboratory can handle. If you are new to how early results relate to eventual medicines, our guide on how to read a biotech breakthrough covers why “works in cells” is only a first step.

The evidence so far: promising, specific and still limited

Tumor organoids and treatment response

Cancer is where organoids have been tested most directly against patients. In a 2018 study in Science, researchers grew patient-derived organoids from people with metastatic colorectal and gastroesophageal cancers who had already received several lines of treatment. The organoids closely resembled the original tumors, and their responses to anticancer drugs in the lab corresponded with how the patients responded in the clinic.

That is an encouraging signal, not a proven clinical tool. The study was relatively small, focused on specific gastrointestinal cancers, and did not show that choosing treatment by organoid test improves survival. Answering that requires prospective trials in which treatment decisions are actually guided by the model. Cancer is many diseases, and a method that works for one tumor type may not transfer to another; our overview of the new cancer toolkit explains why.

To make such models widely available, the US National Cancer Institute and international partners formed the Human Cancer Models Initiative, which produces patient-derived organoids and related models and links each one to clinical and molecular data that researchers can search.

Organ-on-a-chip and liver toxicity

A related technology, the organ-on-a-chip, grows human cells inside small channels through which fluid flows, imitating blood movement and the mechanical forces cells feel in the body. In a 2022 study in Communications Medicine, led by scientists at the chip maker Emulate, a human Liver-Chip was tested blind against 27 drugs with known effects on the human liver. It correctly flagged 87% of the drugs that cause liver injury in people and did not wrongly flag any of the safe ones in that set. It still missed some toxic drugs, and the study involved a company assessing its own product, which is a reason for independent replication rather than dismissal.

The most useful question about any lab-grown model is not “is it human?” but “which human decision does it predict, and how often is it right?”

The rules are changing

Regulation has started to catch up with the science. The FDA Modernization Act 2.0, signed into law in December 2022, rewrote the relevant part of US drug law. Where the law had referred to preclinical tests “including tests on animals”, the bill text substitutes “nonclinical tests”, defined to include animal tests and also cell-based assays, microphysiological systems (organ chips), and bioprinted or computer models. It did not ban animal testing; it removed a legal presumption that animals come first.

On 10 April 2025, the FDA went further, announcing a plan to phase out animal testing requirements for monoclonal antibodies (lab-made antibodies used as drugs) and other medicines, with a roadmap naming organoids, organ-on-a-chip systems and computational models among the alternatives. In April 2026 the agency said it had met its first-year goals, including draft guidance on reducing the use of monkeys in antibody development and draft guidance on using a “weight of evidence” from newer methods. These are FDA’s own descriptions of its progress; draft guidance is not final policy.

Printed tissues are part of the same shift. Our feature on bioprinting human tissue looks at that frontier, and our sister site Bioprinting World tracks it in depth.

What organoids still can’t do

Enthusiasm is easy; the gaps are specific. Current organoids and chips commonly face these limits:

• Missing systems. Most lack working blood vessels, nerves and a full immune system, so they can miss effects that depend on those, such as immune-driven side effects.
• Immaturity. Many organoids resemble developing rather than adult tissue, which matters for diseases of aging.
• Variability. Two batches grown from the same cells can differ, making standardization and regulatory comparison harder.
• No whole body. A single organ model cannot show how a drug is absorbed, broken down by the liver, then acts on the heart. Linked multi-organ chips try, but remain early.
• Validation. For each use, someone must show, against real human outcomes, how often the model gets the answer right.

Brain organoids also raise ethical questions, including about consent for donated cells and how complex such models should become, which researchers and ethicists are actively discussing.

So, could they improve how we test treatments?

Probably, in targeted ways first. The most credible near-term roles are specific, well-validated questions: does this drug damage human liver cells, how does this tumor respond to a panel of drugs, what goes wrong in a rare genetic disease that has no good animal model. Replacing animals across entire drug programs, or replacing human trials, is a much larger claim, and nothing here suggests human trials become unnecessary. Our explainer inside a clinical trial shows why people remain the final test.

Want to see how cells build tissues in the first place? Try our interactive Build a Cell.

What to watch next

• Whether FDA draft guidance on newer testing methods is finalized, and how often sponsors actually use it.
• Prospective trials testing whether organoid-guided treatment choices improve patient outcomes.
• Independent validation studies of organ chips from groups without a commercial stake.
• Progress on adding blood vessels and immune cells to organoids.