These Gene-Edited Donor Islets Kept Making Insulin for 14 Months

Subtitle: Uppsala University Hospital and Sana Biotechnology’s first-in-human UP421 case in NEJM — plus a 14-month update — shows CRISPR-Cas12b hypoimmune islets surviving in forearm muscle with no immune drugs, and still secreting glucose-responsive C-peptide.

Type 1 diabetes is not a shortage of insulin vials. It is a shortage of living beta cells that sense sugar and release the right hormone at the right time. Exogenous insulin saves lives; it does not restore the closed loop the pancreas used to run. The closest clinical stand-in — transplanting donor islets — works, until the recipient’s immune system notices the strangers. Then the only way to keep the graft alive has been lifelong immunosuppression, with its infections, cancers, kidney hits, and the bitter calculus that for most people daily insulin is still safer than the drugs that protect someone else’s cells.

On August 4, 2025, The New England Journal of Medicine published the first human proof that another strategy is possible: edit the graft so the immune system never sees it. Per-Ola Carlsson and colleagues at Uppsala University Hospital, with Oslo University Hospital manufacturing and Sana Biotechnology’s hypoimmune platform, transplanted gene-edited allogeneic islet cells into a man with long-standing type 1 diabetes — without any immunosuppressive drugs. Twelve weeks later the cells were still there, still secreting insulin on demand, and still invisible to the assays that caught the unedited cells in the same syringe. A March 13, 2026 Sana update stretched that story to 14 months.

The paper — “Survival of Transplanted Allogeneic Beta Cells with No Immunosuppression” (DOI 10.1056/NEJMoa2503822) — is a single-participant, first-in-human case report. That is exactly why Sunday audiences should not oversell it. It is also why the numbers inside it travel: they are concrete, checkable, and aimed at the problem every islet transplant has failed to escape.

The patient who had zero of his own

The participant was a 42-year-old man with a 37-year history of type 1 diabetes. Baseline: glycated hemoglobin 10.9% (96 mmol/mol), undetectable C-peptide, daily insulin of 32 units, and autoantibodies (GAD and IA-2) that still advertised autoimmune history. No measurable endogenous insulin. Nowhere to hide a placebo story.

Donor material came from a blood-type–matched 60-year-old pancreas, islets isolated at Uppsala after just over five hours of cold ischemia — glucose-stimulated insulin secretion index 16.7, purity 87%. The islets went to Oslo for engineering into the product labeled UP421.

Three edits, one disappearing act

The hypoimmune recipe is deliberately blunt:

CRISPR–Cas12b inactivation of B2M — strip a key piece of HLA class I.
CRISPR–Cas12b inactivation of CIITA — shut down HLA class II transcription.
Lentiviral overexpression of CD47 — raise the “don’t eat me” signal that tells natural killer cells and macrophages to stand down when they sense missing-self.

Flow cytometry on the final product: 85.8% HLA class I–negative, 100% HLA class II–negative, 46.4% high CD47. The mix still contained some wild-type and double-knockout cells — a built-in control. About 66% of the cells were beta cells; editing did not scramble islet composition.

Under general anesthesia, surgeons made a 3.8 cm incision over the left brachioradialis and delivered 79.6 million engineered cells in 17 intramuscular injections — a “pearl-on-a-string” pattern between muscle fibers. No glucocorticoids. No calcineurin inhibitors. No anti-inflammatories for graft protection. Overnight observation, discharge the next day. ClinicalTrials.gov NCT06239636; funding from the Leona M. and Harry B. Helmsley Charitable Trust.

The immune system attacked everything except the target

This is the part that reads like a controlled experiment living inside one arm. Over 12 weeks, the participant’s blood was challenged against each cell phenotype left in the graft:

Wild-type cells: strong T-cell activation peaking around day 7, IgM then IgG antibodies, complement- and antibody-dependent killing — classic allo-rejection.
Double-knockout (HLA stripped, no extra CD47): mauled by innate effectors — NK cells and macrophages that notice missing-self.
HIP cells (HLA stripped and high CD47): no measurable T-cell activation, no antibodies, no killing in the comprehensive PBMC-plus-serum assay.

The immune system was not asleep. It was busy destroying the decoys. The fully engineered cells never lit up the board.

Function, not just survival

High-sensitivity C-peptide stayed detectable and roughly stable from day 7 through week 12 — near or above ~10 pmol/L after a lifetime of zero. Mixed-meal tolerance tests at 4, 8, and 12 weeks showed C-peptide rising after a liquid meal; baseline MMTT had been flat. MRI at 4 and 8 weeks showed punctate graft signals without inflammation. At 12 weeks, PET-MRI with a GLP-1 receptor tracer ([⁶⁸Ga]Ga-NODAGA-exendin-4) lit up the injection sites — vascularized, living endocrine tissue in forearm muscle.

Four adverse events; none serious, none attributed to the study product. Mild thrombophlebitis at an IV site; local paresthesia possibly related to surgery.

The 7% honesty clause

The shareable miracle has a shareable limit. The 79.6 million–cell dose was intentionally low — roughly 7.1% of the islet mass historically associated with insulin independence (~18 million cells per kilogram in the paper’s framing). Scientific American’s coverage put the same point in plain English: the graft supplied on the order of ~7% of the insulin a person needs to stop shots. Exogenous insulin was actually increased early to protect the fresh graft from hyperglycemia; HbA1c fell by about 42% over 12 weeks, driven largely by that tighter exogenous control, not by a miracle dose.

Carlsson’s SciAm line still lands: “This is the most exciting moment of my scientific career.” Independent voices (Weill Cornell’s Laura Alonso; Yale’s Kevan Herold) called it amazing and preliminary — one person, short early window, clinical significance still unproven until dose and durability scale.

Fourteen months later

On March 13, 2026, Sana reported 14-month follow-up from the same Uppsala investigator-sponsored study: circulating C-peptide still present, still rising on mixed-meal testing, fasting and stimulated levels at month 14 comparable to the first six months and better than a mid-follow-up dip tied to poorer glycemic adherence. 52-week PET-MRI still showed islet signal at the forearm site. No new safety flags in the company readout. The NEJM case was never “cured at three months.” The update is the harder claim: engineered invisibility held for more than a year.

What this is — and what it is not

It is not a license to stop insulin. It is not a population cure rate. It is not Vertex-style stem-cell islets under immunosuppression, and it is not the HIV bNAb remission story — different disease, different tool, same Sunday theme of rewriting the immune bargain.

What it is: peer-reviewed first-in-human evidence that hypoimmune primary islets can engraft in muscle, evade adaptive and innate attack without drugs, and keep making glucose-responsive insulin for months that became more than a year. Nature’s Heidi Ledford framed the September 2025 news peg cleanly; the NEJM primary and the 2026 durability update are the sources that matter.

Next moves write themselves: raise the dose toward independence, move from scarce cadaveric islets to scalable stem-cell sources with the same edits, watch sensitization risk from residual wild-type cells, and keep the long safety ledger that gene-edited non-dividing islet products still need. Carlsson’s own horizon — hypoimmune edits plus a renewable cell source — is the sentence that turns a forearm case report into a pipeline.

Why tonight’s shareable fact sticks

For a century, type 1 care has been replace the hormone. For three decades, transplant care has been replace the cells and suppress the host. UP421 asks a third question: what if you edit the cells until the host has nothing to suppress?

One man. Seventeen injections. Zero immunosuppressants. C-peptide from nowhere. An immune assay that ate everything except the engineered phenotype. Fourteen months of signal still on the scan. Not a pharmacy shelf tomorrow. A credible rewrite of the transplant bargain — and a Sunday story you can tell without inventing a cure press conference.

Sources

Carlsson, P.-O., Hu, X., Scholz, H., et al. (2025). Survival of transplanted allogeneic beta cells with no immunosuppression. The New England Journal of Medicine, 393, 887–894. https://doi.org/10.1056/NEJMoa2503822
Ledford, H. (2025, September 5). Hope for diabetes: CRISPR-edited cells pump out insulin in a person — and evade immune detection. Nature. https://doi.org/10.1038/d41586-025-02802-5
Basilio, H. (2025, August 28). Type 1 diabetes patient’s insulin production restored with new cell transplant therapy. Scientific American. https://www.scientificamerican.com/article/type-1-diabetes-patients-insulin-production-restored-with-new-cell/
Sana Biotechnology. (2026, March 13). Continued positive clinical results through 14 months from type 1 diabetes study of islet cell transplantation without immunosuppression. Company / Nasdaq press release.
ClinicalTrials.gov. NCT06239636 — First-in-human safety study of hypoimmune pancreatic islet transplantation in adult subjects with type 1 diabetes.