Published March 20, 2026 in Nature Biotechnology, from teams at Great Ormond Street Hospital and University College London. The title is measured, the finding isn't: they created a lab-grown esophageal segment that restored swallowing function in pigs without immunosuppression.
Here's the procedure. You take a donor esophagus and strip out every cell — a process called decellularization. The result is a scaffold: extracellular matrix (ECM), mostly collagen, with no cells, no DNA, no biological markers of the original donor. It looks ghostly. It has no living tissue. You're left with the shape and the chemical architecture.
Then you inject the recipient's own cells into that scaffold. In this case, muscle cells harvested from a small biopsy, multiplied in a lab. Two months of culture. The cells populate the scaffold, guided by the positional signals already built into the matrix structure. Then you implant it.
Eight pigs. Three months later, the engineered tissue had contracted, generating peristaltic movements — coordinated muscular waves strong enough to move food toward the stomach. Nerves had regenerated. Blood vessels had grown in. The animals could swallow normally. Their immune systems didn't reject it because the cells were their own.
---
What I keep returning to is the scaffold itself.
The extracellular matrix isn't just a structural container. It carries information. The matrix is organized differently in different regions of the organ — the composition of collagen types, the density of proteoglycans, the alignment of fibers, the concentration of growth factors and signaling molecules embedded in the matrix. These signals are positional: they tell a cell where in the organ it is, what its neighbors should be, what it should become.
This is how embryological development works. The cells don't individually decide what they're becoming through some internal program alone — they read the environment. The scaffold they're embedded in guides them. In development, that scaffold is built by the earliest cells. In the lab, the scaffold was preserved from a donor after the cells were removed.
The question this raises: what exactly is an organ? The naive answer is that an organ is its cells. The esophagus is its epithelial lining, its smooth muscle layers, its nerves, its connective tissue. But strip all of that out and what remains — the matrix — is apparently enough to rebuild the whole thing from a different cell population.
The structure is more than a container. It's a program.
---
The medical context here is long-gap esophageal atresia (LGEA). Children born with this condition have a missing or very short esophagus — the tube that connects throat to stomach is either absent or too short to join. Currently, treatment requires multiple surgeries often spanning years, external gastric tubes for feeding, and frequently complications that persist into adulthood. Some children need transplants. The current standard of care is painful, drawn-out, and imperfect.
Decellularized scaffolds offer a different path: an off-the-shelf donor scaffold, repopulated with the child's own cells. The scaffold can be prepared ahead of time. The two-month culture window is compatible with the current treatment timeline for LGEA. There's no rejection because the cells are the patient's. The scaffold itself is biologically inert after decellularization — no donor antigens to trigger an immune response.
The jump from pig models to human pediatric surgery is not trivial. The team acknowledges this. But the feasibility demonstration in a growing animal — one whose body is still developing, whose repair mechanisms are still scaling — is the most relevant model they could use. The engraftment worked without immunosuppression in a growth context. That's the hard part.
---
I found myself making connections I hadn't planned.
I've been thinking about AI labor displacement, about what gets preserved when a job changes, about what the scaffold of a role is versus the cells that fill it. When AI replaces the tasks inside a job, what remains? Sometimes the role structure persists — the title, the authority, the organizational position — while the actual work is reshaped. Sometimes the structure collapses entirely when the tasks leave. The distinction between "the scaffold of a role" and the cells that filled it might actually be useful.
A job that loses its routine tasks but keeps its judgment calls and stakeholder relationships might be like a decellularized scaffold. The structure remains. New competencies fill it. The position persists.
A job where the tasks ARE the job — where the role is the execution of discrete, repeatable steps — might not have a scaffold that outlasts the cells. Strip out the tasks and there's no matrix left to guide what replaces them.
I'm not sure this maps perfectly. But the scaffold question — what persists when you remove everything active? — seems worth asking about roles, organizations, maybe institutions. What's the extracellular matrix of a company? The culture, the mission, the relationships, the implicit knowledge about how to get things done? Or does the scaffold only hold if the cells that built it haven't left yet?
---
What I can't answer yet: whether the scaffold's positional information degrades. The donor organ was stripped of cells and stored. How long does the matrix retain its signaling capacity? Does it matter how the original organ was used? Does disease in the donor affect the scaffold quality?
These are questions the team didn't fully answer, and they're honest about that. The pig models were healthy animals with healthy donor tissue. Clinical use will be more complicated.
But the core finding stands: the structure knew. Even after everything living was removed, the shape carried enough information to tell new cells what to become. That's not a metaphor. That's the biochemistry.