The Quest to Cure Type 1 Diabetes

THE WEBINAR AND ITS PARTICIPANTS

On Thursday, June 25, 2026, the Kitalys Institute convened a panel of preeminent researchers and clinicians on “The Quest to Cure Type 1 Diabetes.” The webinar drew a global audience via Zoom, with live Q&A and chat running throughout the formal session, followed by a candid 30-minute post-panel discussion. Both sessions were recorded and are publicly available on YouTube (links at the bottom).

The panelists:

• Dr. Alexander (Zan) Fleming — Co-moderator; founder of the Kitalys Institute and Kinexum; former FDA official who oversaw the first insulin analogs, GLP-1 agonists, and metformin.

• Dr. Jay S. Skyler — Co-moderator; Professor of Medicine, Pediatrics, and Psychology at the University of Miami; past president of the American Diabetes Association; longtime leader of T1D prevention research, including the Diabetes Prevention Trial (DPT-1).

• Dr. Chantal Mathieu — Chair of Endocrinology, Catholic University of Leuven, Belgium; leader of the INNODIA and EDENT1FI European collaborative initiatives; most recent past president of the European Association for the Study of Diabetes and most recent recipient of the ADA International Prize.

• Dr. Susan Bonner-Weir — Senior Investigator, Joslin Diabetes Center; Professor of Medicine, Harvard Medical School; a pioneering figure in islet biology and beta cell regeneration.

• Dr. Peter Senior — Professor and Director, Alberta Diabetes Institute, University of Alberta, Edmonton; a leader in islet transplantation.

• Dr. Greg Forlenza — Associate Professor, Barbara Davis Center for Diabetes, University of Colorado; a pediatric endocrinologist running trials of fully closed-loop automated insulin delivery and of verapamil for new-onset T1D.

The timing was notable: days before the webinar, the FDA had expanded teplizumab’s approval — the first disease-modifying therapy for T1D, marketed as Tzield — to include newly diagnosed Stage 3 patients, in addition to its existing Stage 2 indication. And Zan, Chantal, and Jay had recently co-authored a perspective article in Diabetes Care advocating for C-peptide as a regulatory endpoint for T1D disease-modifying therapies.

FRAMING THE QUEST

JAY SKYLER

Jay opened by putting the ambition in historical context — and noting how long that context stretches. He recalled a 1973 ADA postgraduate course at which the slides already featured headlines about artificial pancreas and islet transplantation as imminent cures. More than 50 years later, those headlines still accurately describe work in progress. “You take three steps forward,” he said, “and then two steps backward.” Progress is real but not linear, and each advance tends to reveal new complexities.

He organized the field around four strategies:

• Mechanical beta cell function: fully automated insulin delivery — a device solution that controls glucose but does not address the underlying disease.

• Preservation of beta cell mass and function: immune intervention to arrest the autoimmune destruction of beta cells. Jay added a nuance: the beta cell may not be purely a victim — it may participate in its own destruction, making the biology more complex than originally thought.

• Regeneration of beta cells: stimulating the pancreas to produce new beta cells — through replication, neogenesis from ductal cells, or transdifferentiation.

• Replacement: transplanting beta cells from cadaver donors or, increasingly, manufactured stem cell-derived products.

Jay closed his introduction with characteristic candor: “Can I give you a timeframe? No. My hope is that I will live to eradicate type one diabetes, and that we will get to this before my demise.”

IMMUNOTHERAPY

CHANTAL MATHIEU

THE RECLASSIFICATION OF T1D

Chantal described how dramatically the field’s understanding has changed in the past 20–25 years — particularly regarding when the disease actually starts. T1D used to seem sudden: symptoms appeared, insulin was required, that was the disease. The new picture is very different.

T1D is now understood to begin months or years before a patient ever needs insulin. The field has adopted a three-stage classification:

• Stage 1: Autoimmune attack is underway — detectable by autoantibodies in the blood (against insulin, GAD, Zinc Transporter 8, etc.) — but glucose levels are still normal. The pancreas is compensating.

• Stage 2: Autoantibodies present, and enough beta cell mass has been destroyed that dysglycemia is measurable — subtle glucose abnormalities often only detectable by continuous glucose monitors, such as slight post-meal elevations or early-morning dips. The person has no symptoms and does not yet need insulin.

• Stage 3: Clinical diagnosis — symptomatic hyperglycemia, insulin required.

This reclassification matters enormously for treatment strategy. By Stage 3, only 20–30% of beta cell mass remains. Intervening earlier — at Stage 1 or 2 — offers far more to protect.

TEPLIZUMAB

A FOOT IN THE DOOR

The field’s first approved disease-modifying therapy, teplizumab (Tzield), is a monoclonal antibody targeting specific CD3-positive T-lymphocytes that drive beta cell destruction. It was already approved for Stage 2 T1D (delaying progression to clinical disease), and — announced just days before this webinar — has now been approved for newly diagnosed Stage 3 patients, where it can slow continued beta cell loss.

Chantal was careful not to oversell it: teplizumab is a start, not a cure. It requires infusion, premedication, carries a new FDA black box warning, and is costly. But it is proof that the disease can be modified. “From other diseases like rheumatoid arthritis,” she said, “we know that once you have a foot in the door, there’s a whole world that will open.”

Teplizumab is not the only agent showing promise. The autoimmune attack involves cytotoxic T-cells, inflammatory cytokines, neutrophils, and macrophages, so interventions must be equally diverse. Chantal noted that ultra-low doses of anti-thymocyte globulin (ATG) — historically used in heavy doses to prevent organ transplant rejection — have also shown the ability to disrupt the immune attack in T1D, studied in the MELD-ATG trial. And agents like verapamil and liraglutide work differently still: rather than suppressing immune cells, they fortify the beta cells themselves, reducing cellular stress and helping them survive the inflammatory onslaught.

THE COMPLEXITY OF IMMUNE TOLERANCE

Zan raised the challenge of antigen-specific approaches — using insulin itself, or similar antigens, to induce immune tolerance. Chantal was frank: trying to deploy a tolerizing antigen against an attack that is already “heated up” is like trying to stop a kidney rejection by modifying the antigen during active rejection. Antigen-specific approaches may work very early in life, before autoantibodies are even present — this is the approach being tested by Anette Ziegler. But the evidence for antigen-specific intervention alone, once the attack is underway, is not strong.

Her bottom line: “A lot is happening. Things are moving very, very fast — but not fast enough.”

BETA CELL REGENERATION

SUSAN BONNER-WEIR

Susan opened with a point that reappeared throughout the session: “You’re not going to get anywhere with regeneration in type one diabetes if you don’t have the immune system tamed.” Regeneration and immunology are not competing strategies — they require each other.

She also raised a caution that has set the field back before: what works in a rodent pancreas does not reliably translate to humans. Researchers have successfully induced alpha cells or acinar cells to transdifferentiate into beta cells in mice using viral vectors or chemical injury. Human pancreatic architecture and cellular plasticity are considerably more resistant to such manipulations, and results that look definitive in animal models have repeatedly failed to replicate in human tissue.

THE BIOLOGICAL CASE FOR REGENERATION

Despite that caution, the human pancreas has more resilience than was once thought. Several lines of evidence support the idea that beta cells can be generated anew in adults:

• The pancreas grows nearly 50-fold in volume from birth to young adulthood. The proportion of beta cells within the pancreas stays roughly constant through this growth, meaning the absolute number must be increasing substantially. That growth requires new cell formation.

• The Joslin Medalist Study examined pancreatic tissue donated after death from individuals who had lived with T1D for 50 or more years — the average duration among donors was 67 years. Every single donor still had insulin-positive cells. About 22% had entire lobes of the pancreas with significant beta cell clusters. Susan argues this is not residual cells surviving from early life, but evidence of ongoing turnover: new cells forming, being destroyed by the continuing autoimmune attack, and forming again. Since it is biologically implausible that a single beta cell survives 67 years of continuous immune attack, the pancreas must be continuously attempting to heal itself.

• In longitudinal follow-up of Medalists, roughly 10–15% of individuals who were negative for autoantibodies or C-peptide turned positive when retested 3–5 years later. People with 60+ years of T1D are still generating new beta cells.

NEOGENESIS, REPLICATION, AND PROMISING RESULTS

There are two primary biological pathways to increase beta cell mass in an adult: replication of existing beta cells, and neogenesis — the formation of entirely new beta cells from non-beta-cell precursors. Susan has focused primarily on neogenesis from ductal cells: progenitor cells that can dedifferentiate and then re-differentiate into beta cells under the right conditions.

A notable recent result from Joslin (Peng Yi and colleagues): researchers screened for genes that, when knocked down, trigger insulin expression in human ductal cells. One candidate was Aldh3b2. Using CRISPR/Cas9 to delete this gene in purified primary human ductal cells and transplanting those cells into diabetic mice, they achieved human insulin responsiveness within three weeks. By six weeks, blood glucose was nearly normal — with only 15% of the transplanted cells actually expressing the full identity markers of a beta cell. Susan described it as “very exciting,” while acknowledging the path to a clinical trial is still long.

On the replication side: harmine, a naturally occurring compound, can increase beta cell replication 10-fold in vitro and in human islets transplanted into mice, via inhibition of the DYRK1A kinase pathway. Harmine itself cannot be used clinically — it crosses the blood-brain barrier and causes significant neurological effects — but the DYRK1A mechanism has inspired a pipeline of targeted analogs now entering early clinical trials.

Her refrain: multiple things need to come together — neogenic factors, replication stimulants, and keeping the autoimmunity controlled.

BETA CELL REPLACEMENT

PETER SENIOR

Peter has spent more than 20 years working with islet transplantation, a field that has moved — in his phrase — from “science fiction to science fact.”

CADAVER DONOR ISLETS

THE SUPPLY AND EFFICIENCY PROBLEM

The clinical case for islet transplantation is strong: for patients with severe hypoglycemia unawareness, it can be transformative, giving people back the ability to sleep safely, to be spontaneous, to live without fear. The Edmonton Protocol, developed at Peter’s own institution, first demonstrated that infusing donor islets into a patient’s liver could restore insulin independence. But deceased-donor transplantation runs into two hard limits.

First, there are nowhere near enough donors to treat the millions of people with T1D. Second, the process is inefficient: the biological process of death combined with the mechanical and enzymatic trauma of harvesting is highly destructive to cells — roughly half the islets in a donor pancreas can be extracted, and roughly half of those are lost after implantation in the liver. Most patients need two or more donor infusions to achieve insulin independence. “Those islets,” Peter noted, “are secondhand. They’ve lived a life already.”

STEM CELL-DERIVED BETA CELLS

A NEW ERA

The revolution is manufactured cells. The recipe for guiding embryonic stem cells — or adult cells reprogrammed to pluripotency — through differentiation into functional beta cells has now been worked out with sufficient precision that stem cell–derived islet products are in Phase 3 clinical trials.

These “brand new baby cells” have notable advantages: they come with a full lifetime ahead of them, they can be produced from a standardized cell line in reproducible quantities, and a single dose appears sufficient to achieve insulin independence — unlike the multiple infusions typically needed with cadaver islets.

There is a tantalizing possibility of personalized cells: take cells from a person with T1D, reprogram them to pluripotency, differentiate them into beta cells, and return them — no foreign tissue, theoretically no rejection. This has been demonstrated in a small number of patients in China who already required immunosuppression for other organ transplants, and the results were striking: those patients discontinued insulin. But when the same approach was attempted without immunosuppression, the T1D immune system rejected the cells. “It really, really doesn’t like beta cells,” Peter said. “It doesn’t care whose they are.”

The path forward likely involves a spectrum of immune modulation — not full transplant-grade immunosuppression, but potentially something closer to drugs people take for psoriasis or rheumatoid arthritis: real medications, but manageable as part of daily life.

DIABETES TECHNOLOGY

GREG FORLENZA

THE SEWING MACHINE ANALOGY

Greg acknowledged that he felt like the odd duck in the group — “and here I am, talking to you about computers.” He offered an analogy to explain why this is not as strange as it seems.

During the Industrial Revolution, engineers trying to build a sewing machine kept failing. The reason: they were trying to replicate the natural human sewing motion — passing a needle all the way through fabric, grabbing it, turning it, pushing back. That process was essentially impossible to automate with the mechanical parts of the time.

The breakthrough came when inventors abandoned that approach entirely. They inverted the needle (putting the thread in the tip rather than the heel), added a second thread running perpendicular, and turned part of the device upside down. It does not replicate how a human sews. It is counterintuitive. And it is the basis of the clothing worn by everyone at this webinar.

Automated insulin delivery works similarly: it doesn’t try to replicate the biology of a healthy pancreas. It uses sensors, algorithms, and pumps to achieve glucose control through an entirely non-biological mechanism. And it works.

WHAT TECHNOLOGY HAS ACHIEVED

Greg’s data from his pediatric clinic in Colorado: 80% of patients with T1D are now using automated insulin delivery systems — across diverse racial, ethnic, and socioeconomic backgrounds, including patients on public insurance. Their average HbA1c is 7.8%, down from about 9.5% a decade ago. He describes this as the first generation of the technology.

Fully closed-loop systems — requiring no manual bolusing and having no tunable parameters beyond a glucose target — are now in FDA approval trials. Greg ran these at altitude camp (9,500 feet, 8 hours of exercise daily, irregular schedules, camp food) and achieved more than 70% time-in-range with no bolusing. His prediction: “We will see fully closed loop systems on the market in the United States in 2027 and 2028. It’s not that far away.”

THE HIGH BAR — AND ITS LIMITS

All of this creates a problem for biological cures. Zan put it plainly: technology’s success is a “dark lining” — it sets a bar that is increasingly difficult to clear in clinical trials. If patients maintain excellent glucose control with a device, proving that a biological therapy improves their A1c becomes very hard, and potentially unethical (how do you run a trial requiring people to go off devices that work?).

There is also a physiological ceiling: subcutaneous insulin delivery is inherently slower and less precise than the portal vein delivery of a real pancreas. Technology is very good at approximating what the body does naturally — it cannot fully replicate it.

But Greg was equally clear about what technology cannot do at all: “I can’t make it so you don’t have to wear something.” He described a patient he has treated for 12 years, getting married the following week: “She said, ‘I really wish I didn’t have to wear a pump and a sensor in my wedding dress.’ And I can’t do that.”

He recounted another moment from early in his technology work, when a teenager he was equipping said, “Wow, I feel like I’m turning into a cyborg.” Greg’s instinctive response was “Awesome” — he saw it as a compliment to the technology. The teenager meant it as a negative. Both reactions contain something true.

His conclusion on the relationship between technology and biological approaches: “I don’t think these things are actually competitive. I think they’re synergistic.” Devices can support islet engraftment; devices can give the pancreas rest while other therapies do their work. The goal for biological cures is to clear the bar devices have set — and to offer what devices cannot: freedom from wearing anything at all.

THE PANEL DISCUSSION

REGULATORY, ECONOMIC, AND SCIENTIFIC CHALLENGES

C-PEPTIDE AS THE REGULATORY ENDPOINT

A major thread in the discussion concerned how to measure success in T1D trials — particularly given the technological floor Greg described. Zan, Chantal, and Jay recently co-authored a perspective article in Diabetes Care advocating for stimulated C-peptide as the primary regulatory endpoint for disease-modifying therapies in new-onset T1D.

The argument: T1D is a disease of the beta cell. If a therapy preserves beta cell function, that should be demonstrable by C-peptide — a byproduct of insulin production made in direct proportion to how much insulin the pancreas is producing. Requiring a trial to show improvement in HbA1c, in a world where devices already tightly manage blood glucose, is asking for something technology has made essentially unmeasurable. A drug that preserves functioning beta cells is doing something meaningful, even if the A1c doesn’t move because a device was already managing it.

Jay extended the point: C-peptide is relevant not only for immunotherapy trials but for regeneration and replacement strategies as well. Even islet transplant patients who do not achieve full insulin independence show clinical benefit if they maintain C-peptide — specifically, they eliminate severe hypoglycemic episodes.

Chantal described regulatory progress: EMA has been receptive to these arguments, and the EDENT1FI platform has introduced master protocols for Stage 1 and Stage 2 trials with C-peptide and metabolic parameters as endpoints. The FDA has also begun accepting CGM measures of hypoglycemia as endpoints in some trials, though the panelists pressed for more.

Jay’s pointed question: “The device division of FDA approves glucose sensing to control delivery of insulin — a potentially lethal hormone — and the drug and biologics divisions won’t consider CGM as an outcome measure in intervention trials. What’s the disconnect in the agency?”

THE NEED FOR COMBINATION THERAPIES — AND THE OBSTACLES

Every panelist agreed: no single intervention will cure T1D. The disease requires simultaneously protecting what beta cell mass remains, regenerating what was lost, and replacing what cannot be recovered — while keeping the immune system from attacking all of the above. Combination therapy is not a hope; it is a logical necessity.

The regulatory and industry environment makes combinations extraordinarily hard to execute. Jay described a particularly instructive failure: in 2012, he obtained an IND for a four-drug combination designed to arrest the adaptive immune response, dampen the innate immune response, stimulate regulatory immunity, and improve beta cell health. The FDA approved the IND. The ethics committee approved the study. Eleven centers enrolled. JDRF and the Diabetes Research Institute Foundation each contributed $3 million. He could not get the drugs — each company was afraid that another’s drug would cause an adverse event and earn the combination a black box warning on their product. The study never happened.

“Maybe I was too ambitious putting four drugs together,” Jay said. “Nowadays, I’m only trying to do two at a time.” He noted that teplizumab’s approval has helped: more companies are now interested in the T1D space, and having more approved agents makes combination trials more feasible than running multiple unapproved agents together under IND.

Chantal underscored the regulatory framework problem: current requirements effectively demand proof of each component’s efficacy independently before combinations can be tested. The MELD-ATG trial’s adaptive design — testing five doses of antithymocyte globulin simultaneously, with a dose-determining committee dropping arms — was cited as a model for how to move faster: one adaptive trial can generate the equivalent of four sequential trials’ worth of dose data.

FUNDING AND THE NEED FOR STRANGE COALITIONS

Jay addressed the funding climate directly: “We have a government funding situation which is a disaster.” The current U.S. environment has seen significant research funding cutbacks, including the interruption of at least one late-stage gestational diabetes trial that was one visit from completing its final outcomes assessment.

He took up a question from a webinar registrant about what an “Operation Warp Speed” for T1D would look like — and then explained what would actually be required to make one work, drawing on a conversation with Alex Azar, the former HHS Secretary who ran the COVID vaccine initiative. Azar’s lesson: knowing what to make and how to make it wasn’t enough. The vaccines also required the Army’s distribution infrastructure. Without it, the vaccines would have sat in warehouses. A scientific push requires operational partners that most people don’t think about until it’s too late.

Then he told the story of how the Special Diabetes Program — which provides $150 million per year for T1D research — actually got funded. The diabetes organizations lobbied. The Senate was unmoved: these are disease advocates, tomorrow it’ll be cancer, et cetera. What changed was a phone call from the president of the AFL-CIO to Senator Tom Harkin that the unions were behind the bill. Harkin, who had been skeptical, was persuaded that this was not just a disease lobby but a broad national coalition. To pass it, he partnered with Senator Max Baucus of Montana, who was seeking funding for Native American health — and the two programs were packaged together. “Strange coalitions,” Jay said. “Sometimes that’s what it takes.”

THE ECONOMICS OF ACCESS

Zan raised the equity question: who will actually be able to afford these therapies? Closed-loop systems are not cheap. Cell therapies will likely be more expensive. Teplizumab is already costly and logistically burdensome. Peter offered a framework: T1D reduces life expectancy by an average of 12 years. If an expensive one-time therapy adds those years back — years of earning, contributing, living — cost-effectiveness can look very different than a one-year trial suggests. “The status quo is not awesome,” he said. “Our patients deserve better, and I’m going to keep leaning in.”

Chantal pushed back on the diabetes community’s tendency to accept a cost-effectiveness framework that oncology would never accept: “I’ve been in panels with cancer specialists and they look at me like I come from a different planet. They never ask this question.” T1D is not a nice disease, she said. Even with the best technology, it is never Christmas. You always have to think. You always have to carry your kit. Someone is always woken up by a sensor alarm. “We’re not there yet.”

THE AFTER-SESSION DISCUSSION

After the formal webinar concluded, speakers and attendees who could stay gathered for roughly 30 minutes of informal discussion — candid, ranging, and in some respects more revealing than the structured session.

AI AND THE CGM ENDPOINT

One forward-looking exchange concerned using CGM data in Stage 1 and Stage 2 trials as a surrogate endpoint, when the glucose abnormalities are still subtle. Chantal made the case that current analytics are “20th-century” applications of a 21st-century technology: pattern-recognition AI applied to continuous glucose curves could detect signals of disease progression far earlier than standard metrics. EDENT1FI is already running this experiment — in work package three, with MiniMed, participants in Stage 1 and Stage 2 are wearing sensors and the data are being analyzed with machine learning. Her hypothesis: even in Stage 1 patients, subtle morning glucose patterns may reveal early beta cell dysfunction long before any standard test would detect it.

Greg’s caution: large data sets are available; long data sets are not. Sensors have only been accurate enough to generate reliable signals for about five years. Going further back produces noise — garbage in, garbage out. Inference from shorter time horizons will have to fill the gap.

Zan added the regulatory reality: even if AI identifies compelling patterns, regulators will require validation — what do these patterns predict? That evidence takes time to accumulate. Jay’s observation about the FDA’s current inconsistency on this point — approving CGM to guide delivery of insulin in devices, while resisting CGM as an outcome measure in drug trials — drew a wry response from Zan: progress is being made, but there is further to go.

PREVENTION TRIALS AND THE SCREENING DILEMMA

Jay argued that Stage 1 and Stage 2 intervention already constitutes prevention — specifically, prevention of Stage 3 — and that trials targeting genetically at-risk individuals before any autoantibodies have appeared represent true primary prevention.

But Chantal identified what Zan called the “Gordian knot”: general population screening. EDENT1FI has now screened more than 120,000 children in Europe. Screening itself is fine; people come in and get a test. A first confirmation test, when a screen is positive? Also fine. The hard part is what comes next: asking an asymptomatic person who has never experienced T1D — and likely knows no one who has — to take an injection every two or four weeks, indefinitely, to prevent a disease they have no frame of reference for. “We will have to be very proactive in informing people what type one is,” Chantal said, “without frightening them.”

Zan’s summary: T1D isn’t perceived as a pain point by 99% of the population. Even people who know someone with T1D don’t see it as a threat to themselves. Competing healthcare priorities crowd it out. That is the Gordian knot.

WHAT IT TAKES TO RUN A COMBINATION THERAPY TRIAL

In response to an invitation for panelists to pose questions to each other, Jay reflected on the combination therapy experience described above and what has changed since 2012. The answer: teplizumab’s approval has made the T1D space commercially interesting. More companies are engaged. More agents are in development. Combination trials are procedurally simpler to propose and fund when all components have existing approvals for other indications. The landscape is better than it was — not good enough, but better.

TAKE-HOME POINTS

For those who attended, these points capture the depth of the discussion. For those who did not, they suggest why the recording is worth watching:

• The four strategies need each other. Every panelist emphasized this, but Susan said it most plainly: regenerate all the beta cells you want — if the immune system isn’t controlled, they will be destroyed. No single modality closes the loop. A cure will require combination.

• Technology has raised the bar for biological approaches — and cannot lower it. Automated insulin delivery systems, already delivering near-normal glucose control across diverse pediatric populations, define what a biological cure must exceed — not just on glycemic metrics, but on freedom from burden. Peter’s shorthand for what matters: being able to “shower naked every day.”

• The staging of T1D changes everything about strategy. Stage 1 and Stage 2 patients have far more beta cell mass to protect than Stage 3 patients. Intervening earlier is both more humane and more scientifically tractable. Teplizumab is now approved for Stage 2 and Stage 3; the clinical and regulatory conversation is actively moving toward Stage 1.

• Regulatory innovation is as important as scientific innovation. Adaptive trial designs, platform trials, C-peptide as endpoint, CGM as surrogate — these determine whether the field can test combination therapies on a timeline that serves patients. The MELD-ATG trial and the EDENT1FI platform protocols are models.

• Funding requires strange coalitions. The AFL-CIO story is not just a good anecdote. It is a lesson in legislative strategy: disease organizations alone cannot move policy. Broad coalitions that bring surprising constituencies — the way a union president calling a senator about a diabetes bill was surprising — are what shift the political calculus.

• The people managing T1D are not managing it easily. Chantal: “It’s never Christmas.” Peter: patients “deserve better.” Greg: he cannot yet tell a child they won’t have to wear technology. The technology is extraordinary. The disease is still a disease.

The recordings of the formal session and the informal post-panel discussion are publicly available on YouTube:

Main webinar: https://www.youtube.com/watch?v=hRAZ4F-5zoI

Post-session chat: https://www.youtube.com/watch?v=V3yCwawCr3o