Table of Contents

Author’s Note
Chapter 1: The Beginning: A Parent
Chapter 2: The Scientist
Chapter 3: The Patient Organization: Cure or Treat
Chapter 4: “Now What Do We Do?”
Chapter 5: The Big Gamble
Chapter 6: The Device Makers
Chapter 7: The Inventor
Chapter 8: The Tide Turns
Chapter 9: The Allies
Chapter 10: The Regulators
Chapter 11: The Hackers
Chapter 12: The Patients
Chapter 13: The Finish Line
Chapter 14: The Aftermath
Postscript

Author’s Note

In April 2001 our younger daughter Emma was diagnosed with T1D at age 14. We quickly found our way to JDRF and dove into working with our local chapter in Washington, DC. We were inspired when we heard there would be a cure in five years. We got excited about islet cell transplants only to realize that they would never be of use to Emma. We got excited about every trial that cured T1D in mice. But as the years dragged on, we watched and helped Emma struggle to manage T1D and we ached for her. We started to drift away from JDRF because we no longer believed that JDRF’s singular focus on a cure would help Emma. What was it doing to keep her healthy until a cure was found? Along the way, we met kindred and similarly frustrated and impatient spirits like Jeffrey Brewer. We heard this young new JDRF scientist Aaron Kowalski speak passionately about a new initiative he was leading called the Artificial Pancreas Project. And we were inspired anew.

In 2023, I wrote a story about the development of the first-ever T1D disease-modifying immune therapy, teplizumab. When I finished, I started thinking about other stories in the T1D field that deserved to be told. I quickly focused on the history of artificial pancreas development as a possibility. I realized that while others have written about it, the definitive history of this remarkable journey had yet to be written. In the pages that follow, I try to fill that void.

While this story focuses on a handful of people who played central roles in securing approval of AP systems, space does not allow me to highlight the instrumental contributions of hundreds of passionate scientists, Breakthrough T1D volunteers, clinical trial participants, industry executives and researchers, regulators, lawmakers, and government scientific agencies. While they may not receive the explicit recognition they deserve, what follows is their story as much as it is the story of those featured. They all helped make history, and helped people like my daughter live safer and healthier lives. Our family owes them a debt of gratitude we can never repay.

This story is based on more than 30 one-on-one interviews lasting a total of more than 40 hours, and review of hundreds of pages of peer-reviewed articles in scientific journals. I am responsible for the content of this story in its entirety. I hope you enjoy reading it.

Douglas Lowenstein
Washington, DC
November, 2024

Chapter 1: The Beginning: A Parent

It was 2002 and Jeffrey Brewer, a successful technology and Internet entrepreneur, was worried and frustrated. His son Sean had been recently diagnosed with type 1 diabetes (T1D). T1D is a burdensome autoimmune disease where a person’s immune system mistakenly destroys the insulin-producing beta cells that live in the pancreas.

“The doctors explained to me Sean would be dependent on a hormone called insulin,” Brewer recalled. “They told me the dangers of not dosing the drug correctly: too little and his blood sugar (or blood glucose) levels would be too high putting him at increased risk for long-term health complications; too much and he was at risk for low blood sugar with the potential of becoming unconscious and even dying if no one’s around to help. I was given some needles, insulin vials, a blood glucose meter1, and literally a handwritten sliding scale to calculate correct insulin doses throughout his day. I couldn’t understand why it was so rudimentary.”

Much of the food we eat contains carbohydrates, a form of sugar, or glucose, that our body converts to fuel. In healthy people, when the body absorbs sugar from carbs, it makes just the right amount of insulin to ensure that blood-sugar levels remain safe. But those with T1D are stripped of their capacity to make insulin, so they must execute a delicate dance where the amount of carbs they consume is precisely balanced by the amount of insulin they dose. Over a typical day, the individual with T1D, or their caregivers, must make dozens of complex calculations daily where a single mistake could be fatal. With the technology at hand when Sean was diagnosed, the task was virtually impossible to get it right for a day, let alone for weeks at a time.

Tall and lanky, Brewer was almost always dressed in khaki slacks, a button-down shirt with a school backpack over his shoulders. He was intense and restless, and a friend said, “you can always hear his mind clicking.” These characteristics explain why he was convinced there had to be a better way to manage T1D. And soon he gravitated to an organization he believed would help his son and others safely manage their disease.

The Juvenile Diabetes Research Foundation (JDRF)2 was a nonprofit founded in the 1970s by a group of parents dedicated to funding research to cure T1D. By the early 2000s, JDRF had poured money toward that goal over a 30-year period and while there were some promising advances no biological cure was in sight3.

As Brewer got more involved, he joined a committee of JDRF volunteers that reviewed grant applications brought forward by the research staff, where he had a chance to interact directly with leading T1D scientists. He and other committee members heard one proposal after another to carry out basic science experiments rather than projects that would directly lead to therapies helping people already living with T1D. His doubts about JDRF’s priorities began to grow.

“I realized very quickly that we weren’t anywhere close to a cure for T1D,” Brewer recalled. “In the meantime, my son was going to need something better than what was commercially available for a long time.” With this realization, Brewer had his mission—to keep Sean and others with T1D healthy so that when a cure arrived, they would be healthy enough to benefit. He also had the tenacity and business experience to pursue this goal.

[1] Blood glucose meters are handheld devices where the user pricks their fingertip with a needle that draws a small spot of blood that is then put on a strip and inserted into the machine to generate a glucose reading. The constant pricking can create scars and can be mildly painful.

[2] The organization was rebranded as JDRF in 2010 and in 2024 was again rebranded as Breakthrough T1D.

[3] Around this time, Canadian Dr. James Shapiro at the Alberta Diabetes Institute pioneered islet cell transplants, a procedure to harvest pancreatic cadavers, remove the islet cells where beta cells lived, and transplant them in people with acute T1D. Some people felt this was a major step toward a cure but numerous practical issues, including a limited supply of pancreatic cadavers and the need for recipients to take a lifetime of immune suppressive drugs with adverse side effects, limited its potential for most people with T1D.

Chapter 2: The Scientist

Aaron Kowalski was also frustrated.

Kowalski’s younger brother Steve had been diagnosed with T1D in 1977, and a few years later, in eighth grade, Aaron was diagnosed. “There was a lady who lived on our street fully blinded by T1D,” he recalled. “When we first met her, she had a cane and then she had a seeing eye dog. She was in her twenties. When I was diagnosed, I was like, ‘I just don’t want to go blind.’” Like Brewer, Kowalski was certain there had to be a better way to manage T1D and in his third year of undergraduate work at Rutgers University, he decided to become a biology major “with the purpose of going to graduate school to work in diabetes.” He would go on to earn his Ph.D in microbiology and molecular genetics, anticipating a career in the pharma industry.  

He started working at a hospital in Newark, NJ, doing research on hypoglycemia (low blood sugar). But he wasn’t happy. “I was looking for post-doctorate jobs in various scientific journals and I saw an ad from JDRF seeking a scientific project manager to work on diabetes complications including hypoglycemia, exactly what I was doing. I literally typed up an email and sent it.”

In September 2004, JDRF hired him as a research scientist to lead its program to find ways to mitigate the long-term complications of T1D ranging from blindness to kidney and cardiovascular disease to neuropathy.

Kowalski knew that a landmark study (the Diabetes Control and Complications Trial, or DCCT) by the National Institute of Diabetes and Digestive and Kidney Disease (NIDDK), a unit of the National Institutes of Health (NIH), had found that people with T1D could reduce the risk of long-term complications by 35-76% through extremely tight blood-sugar control—keeping their levels in a range that largely eliminated dangerous highs and lows. There was just one gigantic problem: achieving the DCCT ideal target blood-sugar levels was virtually impossible to do safely for even the most meticulous person living with T1D using the existing “rudimentary” technology.

“The people in the DCCT who had intensive management had an incredible amount of help during the trial 24/7,” said Judy Fradkin, who would eventually become director at NIDDK’s diabetes division. “Even knowing the huge impact it had on them during the seven-year trial, they absolutely couldn’t maintain the level they had in the DCCT after it ended.”

Kowalski projects energy and optimism. He is articulate, passionate, and determined, and his mind, like Brewer’s races a mile a minute. These traits would eventually lead him to become JDRF’s current CEO (Brewer served as JDRF CEO from mid-2010 to mid-2014). But back then, he was at the bottom of the JDRF scientific food chain and one of his first assignments was to attend a conference of the Diabetes Technology Society in Philadelphia. Prior to the main conference session, Kowalski was stunned to meet a group of scientists and patients working with a device that would continuously monitor blood sugars 24/7, a result that blood glucose meters could not match unless a person pricked their finger a few hundred times a day.

“Continuous glucose monitoring (CGM) was a holy grail,” Kowalski recalled of his encounter with the scientists. “I go to this meeting, and they have a CGM and they’re testing it. It alarms when your blood sugars are low or high, you see your blood sugar number all day long.’ It was the greatest news I’d ever heard. It was amazing.” And it offered the hope of achieving the tight controls recommended by DCCT and adopted by clinical organizations like the American Diabetes Association.

The next day, Kowalski attended the plenary session of the conference, and his mind was still buzzing with excitement about the CGM he had seen the day before. The first speaker was a renown T1D endocrinologist who declared that people with T1D can do just fine using the current tools of glucose meters, injections, and insulin pumps. When the Q&A began, someone in the room rose and said, “Doctor, I give you a lot of credit to stand up in front of a room like this and to say something so incredibly stupid and outrageous. That is the dumbest thing I’ve ever heard.”

The mystery man was Jeffrey Brewer and the notion that the available tools were adequate enraged him. Soon after the event, Brewer and Kowalski talked for the first time. Kowalski later said they were “peas in a pod” with a shared determination to improve management of T1D. Their objective was to persuade JDRF to commit research dollars to development of an artificial pancreas system (APS)4 – a system where a CGM would “talk” in real time to an insulin pump and direct it to dose exactly the right amount of insulin to safely control blood-glucose levels with limited intervention by the user. In short, the idea was to come as close as humanly possible to mimicking the function of the human pancreas by automating insulin delivery and relieving people with T1D from the burden of counting carbs, dosing insulin, and making life and death decisions based on gut and math 24/7.

[4] The term artificial pancreas has become common to describe the systems that were eventually developed. They are also referred to as Automated Insulin Delivery Systems. However, technically, the current devices do not mimic a real pancreas for several reasons, including the fact that the pancreas has other hormones involved in blood sugar regulation besides insulin, and they still require some user intervention. Regardless, in the period covered by this story, the term “artificial pancreas” was commonly used and thus that is the case here.

Chapter 3: The Patient Organization: Cure or Treat

An artificial pancreas was not a new idea. In fact, Dr. Ernst Freidrich Pfeiffer developed a machine called the Biostator in 1974, an insulin pump with intravenous (IV) continuous glucose monitoring and IV insulin infusion. The machine was the size of a refrigerator and was only feasible for hospital settings. Thirty years later, the idea of creating a wearable, user-friendly artificial pancreas remained mostly an idea, perhaps dabbled with by some scientists around the world but not of interest to diabetes technology companies or many others. In fact, a few years before the Brewer-Kowalski team formed, JDRF itself had set an organizational goal to develop an artificial pancreas. But it had not allocated any funds toward this goal mainly because senior staff and volunteer Board leadership believed passionately that diverting resources from curing T1D would break faith with donors and throw the organization off course from its singular mission and identity to cure T1D.

John Brady was a member of the JDRF Executive Committee at the time. His son Phillip had T1D. “The intensity and passion of those who had been involved for years was incredibly high. They did not see room for treatment in our mission. They argued we were founded to find a cure, not to treat the disease.”

Brady disagreed and quickly emerged as a key ally of Brewer and Kowalski. “What Jeffrey wanted made complete sense to me—to keep people alive and healthy until we find a cure. We were losing people, kids dying in bed overnight of low blood sugars. If we could automatically dose insulin and have everybody go to sleep and all wake up, that was an incredible victory.”

Brewer and Brady, along with a handful of other volunteers, worked to educate and persuade other key volunteer leaders. Kowalski was working the inside, trying to persuade his research staff superiors to go all-in on funding development of AP systems. He recalled that his bosses told him to stop ‘stirring things up.’ I said I didn’t care. We have got to do this.”

Early in 2005, and now a member of the JDRF International Board, Brewer “decided to force the issue of the artificial pancreas with the JDRF board in the following way: I made a directed gift of a million dollars that was contingent upon that money going to fund proof of concept research for integration of CGM and insulin pumps and development of algorithms for automated insulin dosing.”

Brewer remembered that “they said my gift will dilute the mission, and industry was already taking care of it. But they knew they didn’t want to turn away the money.”

Looking back at the Board meeting, Brady remembered the drama and tension. “It was not a slam dunk. It was probably the most emotional debate that I ever witnessed in all my time on the board. And it was far from unanimous. There was a significant portion of the board who thought it was wrong. This was not something that had strong widespread support, both as a project, nor as a redefinition of our mission to include treat.”   

Kowalski recalled that some of the most influential Board members opposed accepting Brewer’s donation. “Some were vehemently against it. One of them was on the Board of Johnson & Johnson and he said ‘J&J’s research budget for diabetes is hundreds of millions of dollars. You’ll never make an impact.’ It was a huge throw down. And Rob German, the chairman, said, ‘this guy’s passionate, he’s willing to put his money up. Why don’t we get somebody on the board, set them up with Jeffrey and this Kowalski kid and give them six months to convince us you can spend a million bucks.’” Said Brewer: “They didn’t know partnering me with Aaron was basically putting the fox in the hen house.”

The study commission was led by highly respected Board member Charles Queenan. In October 2005, Queenan presented the study commission recommendations including “to pursue [an artificial pancreas] with the same intensity as other JDRF cure therapeutics goals.” According to Board minutes, the Board agreed to approve the recommendations.

With this, the Artificial Pancreas Project (APP) lifted off. It marked a turning point in JDRF’s research priorities as the organization fully embraced the commitment to keeping people healthy on the path toward a cure. Within a year, the emotional battles over cure vs treat would be largely forgotten. The JDRF commitment would help catalyze the most significant advance in T1D treatment in nearly a century. Between 2005 and 2024, JDRF would spend $171 million on artificial pancreas related research that would result in the commercialization of a series of systems that partially automate insulin delivery and dramatically reduce the daily burden of T1D while improving the daily and long-term health of people with the disease.

Chapter 4: “Now What Do We Do?”

There is a scene in the movie The Candidate where Robert Redford’s character, Bill McKay, has learned he was the upset winner for a U.S. Senate seat in California. Just before McKay goes down to speak to his supporters, he turns to his chief advisor and asks, “What do we do now?” 

That was the exact question facing the small team at JDRF charged with creating an artificial pancreas. The diligence study had made one thing clear: this would not be your mother’s science project. It would have to run on multiple parallel tracks: the science and engineering track—developing algorithms—the mathematical formulas that would track data from the CGM and adjust insulin dosing minute by minute to keep blood sugars in a safe range; the regulatory track to understand what the FDA would require to grant approval of an artificial pancreas; the industry track to encourage investment and commercialization of AP systems; and the reimbursement track to persuade private and public health plans to cover the system upon approval.

It was unlike anything JDRF had ever undertaken. And it would need someone with a unique set of cross-sectional skills to lead. JDRF CEO Peter Van Etten had just the guy: Larry Soler, JDRF’s Washington, DC-based chief lobbyist who had organized and led a multi-faceted advocacy campaign to overturn bans on stem cell research and JDRF’s advocacy campaigns to ensure multiple renewals of the Special Diabetes Program (SDP) that provided NIH with hundreds of millions of dollars for T1D research since its original enactment in 1998. To Soler, his work was more than a job. He himself had been diagnosed with T1D in his early 20s.

And Soler had exactly the right partner in a recent hire who had unexpected bandwidth due to congressional paralysis on an unrelated issue. Her name was Cynthia Rice, a Harvard graduate with experience as a Senate staffer, working in nonprofits, and a stint staffing President Clinton’s Domestic Policy Council specializing in health policy. Soler and Rice were both preternaturally calm. Both were highly meticulous and organized, and both knew their way around the corridors of power in Washington, DC, which would be an epicenter of activity given the presence of the U.S. Food and Administration (FDA), the regulatory body that would have to approve for commercial use in the United States any artificial pancreas system. (Soler would leave JDRF in 2009 to become the first CEO of Michelle Obama’s anti-obesity nonprofit, Partnership for a Healthier America, and Rice would take over the lead role, seeing the project to its completion before she retired in 2023).

As the team organized and started building out a strategic plan in December 2005, the NIH hosted the first of what would become a series of key research and interagency conferences on artificial pancreas technology over the next decade. The focus was on exactly how to pursue the goal. It turned out that not everyone there felt the goal was attainable.

One of the attendees was a University of Virginia Ph.D. mathematician named Boris Kovatchev. Kovatchev’s father had been diagnosed with insulin-dependent diabetes at the age of 48 and it made a huge impression on him. “He had a couple of toes amputated. It was a really bad case.” Kovatchev especially grasped just how difficult it was for people with the disease to correctly dose insulin, and upon arrival at UVA in the 1990s, he quickly started specializing in T1D, including tinkering with algorithms to manage insulin dosing.

“There were debates at the NIH meeting whether it is even possible to do an artificial pancreas,” he recalled. “There was a prominent scientist who said it is not possible. There were essentially two camps, one was ‘let’s try it,’ and the other was ‘it’s not possible.’”  Said Kowalski: “This prominent researcher stood up and said ‘you’re going to kill people.’” But despite some skeptical voices, Kowalski told the gathering that JDRF would move ahead and seek applications to fund key research to advance the project, especially in the area of developing the critical algorithms that would be the brains of the operation.

By August 2006, teams of researchers at multiple academic centers had received grants of about $300,000 each to take on various aspects of AP development. Some had dabbled in the field for years, others more recently. But collectively they were a Murderers’ Row of T1D scientists. Bruce Buckingham at Stanford, Bill Tamborlane and Stu Weinzimer at Yale, Kovatchev at UVA, Roman Hovorka at the Wellcome Trust-MRC Institute of Metabolic Science at the University of Cambridge in the UK, and Frank Doyle, Howard Zisser, and Eyal Dassau at University of California at Santa Barbara (UCSB)5 were among the grant recipients.

Another early grantee was the Jaeb Center for Health Research, founded by Roy Beck. His organization’s role wasn’t the headline-grabbing work of conducting high-profile trials, but it was a critical component of the project. For the next decade, Jaeb coordinated virtually every important clinical trial, engaged with the FDA on trial design, captured and analyzed data from the trial, and supported the publication of the research in peer-reviewed scientific journals. It also served as secretariat for the JDRF-created Artificial Pancreas Consortium, a series of meetings over the years bringing virtually every major researcher, funder, and company in the field together to exchange information and ideas. “What Jaeb did was herd some of the wildest, cockiest people,” said Kowalski. “They helped us enforce the rules of the consortium so people weren’t going off on their own screwing things up. Roy is probably one of the most important people in diabetes that nobody’s ever heard of.”

While researchers were energized, two immediate challenges emerged.

One was the U.S. Food and Drug Administration (FDA), the regulatory agency that reviews and must approve for commercial use the safety and efficacy of all medical devices and drugs. In a meeting in the early months of the project, JDRF’s team met with FDA officials to explore what data and studies the agency would want to see as a basis for reviewing AP systems. The response was discouraging.

“We went into their conference room and, boy, did they shut us down,” recalled Soler. “They basically told us that if we are thinking about testing these systems in humans we should instead be talking about testing in mice. I remember leaving and feeling, whoa, this went really bad.”6

The second concern was the reliability of what would be the nerve center of the AP system—the continuous glucose monitors that Kowalski was so enthused about months earlier. FDA had approved early versions of CGMs in the late 1990s, but they did not actually perform real-time blood-sugar monitoring. In 2006, it approved two newer “real-time” systems developed by the major device maker Medtronic, and a new start-up named Dexcom. But the FDA only allowed these CGMs to be used as a secondary measure of blood-glucose checking because it still felt that the finger prick meters were more reliable.

“FDA didn’t allow the use of CGM data for decision making and could not reconcile that insulin would be dosed automatically based on a reading from a device they deemed inaccurate,” said Sanjoy Dutta, currently chief scientific officer and head of Research at Breakthrough T1D (formerly JDRF) who joined the JDRF APP team a few years after the project started. “The initial CGMs had high levels of inaccuracy,” said Dutta. “They were the Achilles Heel of developing AP systems.” 

FDA’s concerns about CGMs were troubling enough without the fact that health insurers declined to cover them (they could cost up to $5,000) because they were not persuaded that they were more accurate than existing glucose meters, a view reinforced by the FDA’s decision to approve them only as secondary measures. If insurers would not cover CGMs, it was even less likely they would cover even more advanced and complicated AP systems, and without some certainty around coverage the prospect for AP development would be dim.  

[5] Each of these researchers were part of larger teams that participated in the groundbreaking work, but it is impossible to list all the names here.

[6] Typically, research includes a preclinical phase where drugs and devices are tested in animals, often mice. But these studies are costly and lengthy and would set the AP project back years before it got off the ground.

Chapter 5: The Big Gamble

“We realized that if the sensors weren’t successful, we couldn’t get to a closed loop,” said Kowalski. “And nobody believed in the sensors. They all sucked. And we went around and asked ourselves what we need to do” to prove that even less precise early sensors would still be better for patients than glucose meters. The discussions led JDRF to take a big gamble and invest $10 million to conduct a yearlong, multi-center trial of the new sensors now being developed by Medtronic, Dexcom, and a third player, Abbott, to prove to insurers that CGM were more accurate than glucose meters.

“We made all the decisions about this trial, but we consulted with the companies,” Rice said about the team’s meetings and calls with device makers and health insurers. “The companies were very risk averse.” But ultimately, working with Jaeb, and believing they could design a sound study that would prove the latest CGMs were superior to any existing blood meters, JDRF moved forward. The trial that would be historic for its folly or brilliance launched in September 2006.

Kowalski remembers a meeting with Claudia Graham, a senior executive at Medtronic. “She said to me ‘you guys could screw up this whole field if the trial fails; this whole field will be dead.’” Graham remembers the conversation. Her comment was rooted in her concern that the first generation CGMs were “kludgy” and finicky and hard to use for many. Concern about doing the CGM study) was that these were all first-generation devices. “I was very worried the results would reflect those imperfections at the time.”

JDRF’s Soler said: “The trial was a huge gamble. It was hugely expensive, and the goal was to show that people who use CGMs do better than people who don’t.”

If that goal was not achieved, Graham’s fear would be reality.

Chapter 6: The Device Makers

Al Mann was the son of a Portland, OR grocer and a mother who was a pianist. Mann’s brother was a violinist who became a founder of the famed Juilliard School of Music, and his sister was a concert pianist. But Al went another direction, getting a degree in physics and becoming a serial entrepreneur, starting companies in fields ranging from aerospace to cardiac pacemakers. Eventually, in the late 1970s, he turned to diabetes, starting a company called MiniMed which developed insulin pumps. In 2001, he sold MiniMed for $3.7 billion to one of the country’s leading medical device makers, Medtronic. By 2005, Minneapolis-based Medtronic was the undisputed giant in the T1D device space. Al Mann remained involved, and he was the person who flatly stated that at the 2005 NIH meeting that AP systems would “kill” people.

To JDRF, engaging industry was a critical objective because products are not developed and sold by academic scientists working in labs. Indeed, a foundational tenet of the AP Project was to catalyze industry involvement. In fact, in one of Brewer’s presentations to the JDRF Board, he had a slide which simply said: “Without JDRF, still ten years away.” So, JDRF made it a priority to attract Medtronic to AP development. But the company was not taking the bait.

“They were sitting fat and happy owning the insulin pump business,” said Brewer. “We went out to their headquarters in Minnesota for a big dog and pony show. I remember they had an iceberg slide which showed nothing at the top, but they told us what’s going on with AP was confidential and below the surface and that we didn’t have to worry about it, and we should go back to funding the cure. But we knew they in fact they had few people working on it under a small NIH grant and they weren’t putting a dollar of their own capital into the AP. They didn’t see the need to do so. It wasn’t going to help them expand the pump market which they already owned.”

“The problem with Medtronic was there was a certain degree of arrogance,” said Graham who retired from Medtronic in 2008 (she eventually joined Dexcom. “They thought it was all about the pump. They had hundreds of thousands of pumps in the field with four-year warranties so for them to easily shift over [to AP systems] was a big business hurdle.”

Medtronic’s seeming slow pace created a new imperative. Abbott, Dexcom, and Johnson & Johnson’s diabetes subsidiary Animas were much smaller players and fighting to gain a foothold in the CGM market. Leaping directly to AP development was seen as high risk.

“For JDRF the fundamental problem we had was to create competition where there was no meaningful competition in the pump market,” recalled Brady. “Medtronic was the gorilla and had the ability of doing it, the technology, and the resources but there was no competitive market incentive for them to move forward.”

Chapter 7: The Inventor

When Dean Kamen was in his teens growing up in Long Island, NY in the mid 1960s he started making and selling to local bands homemade lighting control systems. Around this time, his older brother Barton had earned a Ph.D. in pharmacology and was focused on pediatric oncology.

“My brother would come home and explain what he was doing in neonatal ICU. He had some ideas on how to do cancer treatments better in babies. He wanted something that could give them very small amounts of drugs and to time the dose in intervals instead of running on an IV drip.” Kamen developed a pump device that met his brother’s specifications, using off-the-shelf components including the base of a standard butter dish. He made the units in his parent’s basement and drove them to his brother at Yale. “It wasn’t a business,” he recalled. “I was just trying to help my brother save babies.”

But the brotherly collaboration would change diabetes history. A few years later, Dean Kamen was at Worcester Polytechnic Institute. His brother Barton showed the pump device to a colleague at Yale. The colleague was the same Bill Tamborlane, who would co-lead some of the most seminal AP trials years later. But at the time, he was working with pregnant women with T1D.

According to Kamen, Tamborlane envisioned adapting the pump system to dose insulin. He imagined a pump that would provide a steady baseline, or basal, dose of insulin through the day, and allow the user to give a higher, or bolus, dose at mealtimes to account for the carbohydrates consumed. Kamen took on this challenge and developed a wearable pump weighing 17 oz. It was the first insulin pump in history. It was known as the “blue brick” because this time Kamen could only find blue butter dishes. Soon, Kamen created his first company, Auto Syringe, with the blue brick at its center. In 1981, the drug company Baxter International purchased Auto Syringe.

The Big Blue Brick

After the sale to Baxter, Kamen continued to create inventions in a range of fields, including healthcare. But he stayed abreast of the insulin delivery and CGM market, and, by 2005, his mind returned to whether he might invent a way to automate insulin delivery combining a CGM with an insulin pump. “I thought if they can make sensors that can measure blood glucose levels every couple of minutes we ought to be able to do a closed loop control so I said, ‘I got to go back and develop a technology that eliminates the syringe and its mechanical limitations, I got to make it small and light, and I’ve got to invent a measuring system that can measure down to fractions of a microliter how much insulin I am delivering every time.’”

So, it made sense that when JDRF began to explore options to develop AP systems, Kamen, by then founder and CEO of DEKA Research and Development Corporation and now world famous for his invention of the Segway, would be one of their first stops.

It soon became clear that the dream of creating competition might be realized.

“We went and toured the engineering facility, and we ended up making multiple visits up there,’ said Soler. And in the process, they learned that Kamen had been exploring with Abbott the possibility of creating an AP system pairing Abbott’s Navigator CGM, with Kamen’s new pump and dosing algorithm that his team developed. Soler recalled the team’s emotions as the dialogue continued. “You had a company that had a very promising CGM. They were building a significant insulin pump with Dean. Dean was an innovator. Dean had built the first insulin pump and has been involved for a long time and had a record of innovation. And so, we got really charged up by it, and we went for it.” “It” in this case was putting together a multi-party deal between Kamen’s DEKA R&D, Abbott, and JDRF.

“We had negotiations about the different IP rights, how much funding we would invest based on milestones, and what our royalty rates would be,” Soler recalled. “It was a long process, but by 2008 we got to the point where we had a deal that was agreed to by the parties.” As the deal wrapped up, Abbott announced that the President of its diabetes division who had been its lead negotiator would move over to become president of the cardiac division. The news did not trigger any alarm bells at JDRF.

“We scheduled an in-person meeting at the JDRF headquarters in New York to sign the final documents,” Soler said picking up the story. “It was in the evening and lower Manhattan was dark. The lights were out in every office except for the one conference room. We had the papers all laid out for signature and were waiting for the new executive in charge, Heather Mason, to arrive. But we kept getting alerted that the company plane was delayed. And finally, she contacted us directly and said she didn’t want to do the deal. So, she never showed up. We thought we would be celebrating. But she left us waiting.”  The deal was dead.

Chapter 8: The Tide Turns

Despite the crushing disappointment over the Abbott deal, there was still cause for hope. For starters, the scientists JDRF funded in 2006 were delivering promising results.

In 2007, Doyle’s team at UCSB had published data showing that they had developed a closed loop system using a Dexcom CGM and an Omnipod insulin pump made by Insulet that communicated with one another using an algorithm that adjusted the insulin infusion rate every five minutes based on predicting future glucose levels. It was an important landmark in AP development.

At UVA, and across the Atlantic in Padua, Italy, Kovatchev and Claudio Cobelli, were collaborating to develop a simulator—a computer model—representing 300 different metabolic systems that allowed them to test scores of real-life scenarios, such as creating scores of different types of meals with varying carbs, different forms of exercise for different periods, and then testing multiple different algorithms to see, among other things, how much time the simulated “people” were in range, what risks emerged, and how many hypoglycemic events occurred. The idea was to prove to the FDA that these simulated models were an accurate substitute for testing artificial pancreas algorithms in humans not animals.

In January 2008 the FDA accepted the simulator as a substitute for animal testing. Three months later, Kovatchev’s team started an in-hospital human trial at UVA. Cobelli and a third colleague, Eric Renard in France, quickly followed suit. “The simulator saved at least five years of animal studies because we didn’t have require an algorithm to be tested in an animal model to be deemed safe and effective before going into human studies,” said JDRF’s Dutta. “That entire chunk was eliminated thanks to Boris and Claudio.”

Components of the first system used at UVA

Also, in 2008, Weinzimer and Tamborlane at Yale published results of a first in human study of 17 adolescents using a rudimentary but nonetheless fully automated closed loop system. They reported that “closed loop glucose control using an external sensor and insulin pump provides a means to achieve near-normal glucose control in youth with T1D during the overnight period.” When the team presented their findings at a JDRF research summit, “people were literally crying,” Weinzimer recalled.

In even bigger news, in September 2008, JDRF’s big gamble to run a CGM trial paid off. The trial proved conclusively that CGMs are associated with improved blood-glucose control in adults with T1D compared to those who used standard therapy. “People still cite it as the landmark JDRF CGM study,” Kowalski said proudly. “I remember when that paper was published in the New England Journal of Medicine that it was like the whole world changed.”

The goal of the paper had been to persuade health insurers to initiate coverage of CGMs. But even before it came out storm clouds formed. One of the largest insurers, United Healthcare, was on the verge of issuing a non-CGM coverage decision. It would have sent a potentially fatal message to other payers, potentially crushing the CGM market before it was off the ground, and even taking down the artificial pancreas market still in its infancy. JDRF tried to reach out to its contacts at the company but hit a brick wall. Cue its network of well-connected volunteers.  

“We learned that United Healthcare was planning a CGM non-coverage decision,” said Rice. “I put out an all-points bulletin to our network, asking if anyone had connections to senior executives at the company. Lo and behold, one of our leaders, Pam Sagan, had been friends for many years with a UHC executive and his wife, and that executive had since become CEO of the company. Pam was able to get through to him quickly and share that important new research was going to be published shortly, and it would be in UHC’s best interest to delay its non-coverage decision to reconsider the new evidence,” Rice recalled.

“The CEO got the decision delayed and a meeting for us to brief senior medical staff on the new findings. United Healthcare—the largest health plan in the country—ended up covering CGMs based on the JDRF study results and this volunteer intervention.” Soon, virtually every insurer would announce a favorable coverage decision.

But there were still some headwinds. Kowalski kept hearing grumbling that the project was going nowhere because there was no obvious commercial product. Sure, the lab research with somewhat jerry-rigged systems was nice but they were too cumbersome and complicated to actually be developed into a viable, wearable commercial device.

Kowalski knew they had a point. Industry needed line of sight into where all this would lead not happy talk and assurances from JDRF and scientists. What emerged would eventually be known simply as The Roadmap, a set of six boxes with iterative devices that suggested concrete products with escalating complexity and automation.

“When we got the CGM trial data, apart from the fact that the sensors helped people, a big ‘ah ha’ was nobody appreciated how much time people were spending with high and low blood sugars every day,” said Kowalski. In fact, the data found that if you were hitting one of the targets recommended by clinical organizations based on the DCCT data, “you were still spending almost 12 hours a day with a high blood sugar and 80 minutes with a dangerously low one. It blew people’s minds.” 

This data crystalized a path forward. “I had the idea that a first product could turn off the pump when people are low” so that it didn’t continue to dose insulin and plunge them into severe hypoglycemia. The so-called low glucose suspend system would not be the ultimate goal for automated insulin delivery, but it would be less daunting to develop. And by suspending glucose when the CGM reached a specific low reading it would help overcome FDA’s concerns about rogue glucose sensors triggering a dangerous low blood-sugar event by continuing to dose insulin well past the safe level.

In the end, Kowalski created six product visions grouped into boxes. They culminated in Box Four, a “hybrid” closed loop where insulin delivery would be mostly automated except at mealtimes when the user would have to “bolus” to account for the carbs consumed (thus hybrid closed loop, not fully closed loop), Box Five where, a fully automated closed loop, and Box Six, a system that not only dosed insulin but other pancreatic hormones that impact glucose control in Box Six. Published in Diabetes Technology and Therapeutics in September 2009, the paper for the first time laid out a manageable development plan for companies that made the ultimate goal seem more attainable. Looking back, Roy Beck said that Kowalski’s seminal paper “has been the roadmap for the last 15 years in artificial pancreas development and is still cited.” 

And in January 2010 JDRF’s efforts to engage industry paid off. It struck a deal with Johnson & Johnson subsidiary Animas to build an AP closed loop system using a CGM provided by Dexcom. JDRF committed $8 million over three years to support the project. Recalling how important this was, Soler said: “We had a major medical device company make a public declaration that they were going to create an automated insulin delivery system, and it really galvanized the attention of the diabetes industry.” (The irony was not lost on some that in the JDRF debate whether to accept Brewer’s funding, one leading opponent had declared that J&J was never going to invest in the artificial pancreas space. Five years later, they did just that.)

Chapter 9: The Allies

David Panzirer’s daughter Morgan was diagnosed with T1D in March 2007. Immediately, his grandmother, New York real estate mogul Leona Helmsley, made a $5 million donation to JDRF and another organization in the T1D field. Upon her death five months later, her will stipulated that her assets be liquidated and transferred to a charitable trust, Helmsley Charitable Trust (HCT). Panzirer became a trustee, and together with his co-trustees they set up a series of core programs including a focus on T1D. If Morgan’s diagnosis wasn’t enough motivation, his other daughter Caroline was diagnosed in 2017. Over the years, Panzirer has developed a visible, respected, and well-cultivated image as someone willing to be outspoken and willing to disrupt the status quo if he feels it would help the lives of his kids and the entire T1D community. HCT and JDRF have partnered and collaborated on numerous transformational initiatives, with occasional friendly disagreements.

HCT would prove to be a powerful ally, lending its voice to the effort to educate the FDA on T1D and tear down its regulatory barriers. One of its major contributions to that effort was to build the first-of-its-kind registry on T1D patients. It started in 2007. “We kept on asking when we started where the data was to show how people in the U.S. are doing with T1D,” Panzirer recalled. “The data did not exist. It was mostly thought of as a safe and managed disease.” About three years later, during JDRF’s campaign to turn the FDA around (described in the next chapter), Panzirer presented the registry data to the agency. “The registry collected data on 3,000 people in 27 clinics and showed that T1D was neither managed nor safe,” said Panzirer. “I presented the data myself and their chins were on the floor.” Said Kowalski, “The registry was very helpful. It was super important to help show the unmet need.” 

HCT also stepped in to beef up JDRF’s team. In 2010, with the work expanding and the possibilities for success growing, HCT provided a transformational grant providing resources for JDRF to hire an expanded team for the AP campaign, including Campbell Hutton, a regulatory expert, scientist Sanjoy Dutta, Linda Johnson, an expert in managing alliances with industry, and Marlon Pragnell, another research scientist. “It was huge,” said Soler. “They became central members of the team.” “HCT provided resources that allowed us to take the JDRF Artificial Pancreas project to the next level,” Kowalski said.

HCT would also become a leading funder of the multi-hormonal “bionic pancreas” that Boston University researcher and T1D parent Ed Damiano was developing. The Kowalski roadmap had included such a system as the final Box Six, the ultimate manifestation of a real pancreas. But JDRF’s research priorities focused in these years around systems it felt were more realistic from a developmental and business standpoint, and HCT filled the gap by investing in Damiano’s project.

In 2012, Panzirer shook things up a bit more. With concerns about CGM sensors continuing, Panzirer approached JDRF with a proposition. “I went to them and said that the sensors are not good enough and that Helmsley is going to put up $12.5 million and I want you guys to match it, and we will do an RFP to companies who can improve sensors,” Panzirer said. Kowalski remembers that initially the idea of potentially providing direct funding to companies like Medtronic did not go over well at JDRF. Kowalski said the idea that JDRF would collaborate with the largest medical device maker in the U.S.” was very controversial. “But in the end, the two combined to provide $17 million to Medtronic to improve sensor technology. Notably, the deals contained royalty rights if the grants resulted in actual commercial products. While those sensor investments did not immediately bear fruit, they would ultimately result in meaningful improvements in Medtronic sensors that are currently coming to market. “To date,” says Panzirer, “we’ve gotten $42 million in royalties back.”

The other force multiplier was NIDDK. Prior to the launch of the AP Project, NIH had for years been providing small grants to researchers dabbling in AP research. But as JDRF stepped up its campaign, NIDDK would join it, leveraging funds from the Special Diabetes Program that JDRF helped create and sustain. Four years into the project, in 2009, it disbursed $23 million in research grants that added major accelerant to the project, and continued NIH funding eventually led directly to algorithms that are being used today in several commercial AP systems.

Chapter 10: The Regulators

Five years into the project, there was cause for optimism. But the FDA remained a major problem, even an obstacle.

From the start of the AP Project in 2005, there was frustration with the position of the FDA towards the technology and its potential. By 2010, despite the approval of Kovatchev’s simulators to replace mice trials, these frustrations were boiling over as FDA was continuing to raise barriers to conducting clinical trials, and generally showing little interest in advancing the technology. In 2010, Brewer became CEO of JDRF. “The FDA had been very unhelpful. There were people there overseeing the field who didn’t think there was any need for technology to manage insulin and glucose. They felt there was already an acceptable treatment and people should do what their doctor directed, take insulin as prescribed, and they would be fine,” recalled Brewer.

JDRF and researchers were pressing FDA to allow outpatient trials not just ones in highly supervised hospital settings. And it wanted the agency to develop an Artificial Pancreas Guidance, essentially a roadmap laying out the steps that needed to be followed to get device approval: what trials are required, how they should be designed and what their endpoints should be, how many patients had to be included and what ages, what technology might be used to capture data like blood-glucose levels, among other key regulatory issues.

“There were multiple years of FDA wanting more and more data before trials could progress, which we saw as over-regulation,” recalled Kowalski. “So, frustration boiled and boiled and boiled, and we finally went to Jeff Shuren who was the new head of the Center for Devices and Radiological Health (CDRH) at the FDA and asked him to get involved.”

But they weren’t getting anywhere, so in July, JDRF brought together a broad set of stakeholders to develop recommendations on how to conduct outpatient trials safely. “In November we presented the recommendations to the FDA and other experts at an FDA/NIH public workshop,” Rice said.

It did not go well, and JDRF decided to turn up the heat. 

“We were so concerned with the dismissive reaction from the lead FDA reviewer at the workshop that the same day, right after the workshop ended, we went to Larry Soler’s house nearby to confer about next steps—including me, Aaron, Jeffrey, Larry, Campbell Hutton (the in-house JDRF regulatory expert), and our regulatory consultant Phil Phillips,” recalled Cynthia Rice. “At that meeting, we decided we would draft an actual guidance document, in the format that the FDA could release, based the earlier recommendations.” This would lay out a pathway to get to the “Holy Grail” of Box Six in the Kowalski roadmap. The guidance document was submitted to the FDA in March 2011.   

Over the next six months, JDRF waged an intensive, multifront advocacy campaign to send a message to FDA that it was time to move faster and more constructively on AP systems. It started in March when volunteers from all over the country flew into Washington, DC for JDRF’s annual Government Day and swept across Capitol Hill asking lawmakers to sign a letter being circulated by the House and Senate Diabetes Caucus Co-Chairs7 to the FDA to promptly issue a guidance that reflects the view of expert clinicians and researchers.

In June, JDRF staged its biennial Children’s Congress to educate lawmakers about T1D. Long-time JDRF ally Senator Susan Colins (R-ME), Chairman of the Senate Aging Committee, agreed to schedule a hearing to press FDA officials about the lack of progress in advancing artificial pancreas systems to patients. The stagecraft would be designed to create an emotional exclamation point on the issue as the 200 delegates between the ages of 5 and 17 sat in chairs and on the floor in front of the dais in the hearing room.

FDA was getting the message and tried to get ahead of things. It released a guidance document the very same day of the hearing. The bad news is the guidance was narrow—it only covered a pathway for a low glucose suspend version of the AP—Kowalski’s Box One, a first-generation device already available and in use in more than 40 countries around the world that researchers already felt they could leap beyond. Not only was it viewed as too conservative, but there was also another problem: said Rice: “It was terrible.” 

In September, joined by key clinician groups like ADA, individual researchers and clinicians, JDRF submitted comments blasting the FDA’s June draft. In October 2011, it launched an online petition calling on FDA to issue a pathway to accelerate AP development. In short order, over 120,000 people signed the petition, most with a direct connection to T1D. And one month later, Senators Collins and Shaheen held a press conference, joined by an 11-year-old with T1D, Caitlin Ryan, urging the FDA to issue clear and reasonable guidance for AP that avoids unnecessary delays. A foot-high stack containing all the names of the petition signers was a prominent prop.

The final salvo came that same day. Full-page ads appeared in The New York Times and Washington Post headlined: “The FDA can help save the lives of those with type 1 diabetes.” Pictured on the top of the ad was an adorable young girl named Piper, an eight-year-old who had been diagnosed at age 3. The ad copy said: “Piper has type 1 diabetes. One in twenty people like Piper will die from low blood sugar. In fact, kids and adults are dying every day from low blood sugars or complications caused by type 1 diabetes. In the next few weeks, FDA has a chance to show it is leading the world in medical innovation not standing in its way. It will lay out the pathway to bring to market the first artificial pancreas, a lifesaving technology now under development. And the most revolutionary treatment in diabetes since the discovery of insulin. Three million kids, teens, and adults with type 1 diabetes are counting on the FDA to get it right. Our lives and health are at stake.”

To this day, people at FDA at the time, JDRF staff, and volunteers, remember “the ad.” It remains the single most public and aggressive action the organization has ever taken to advance research to treat and cure T1D.

In December, FDA released a second draft guidance on outpatient testing of AP systems. But it too didn’t meet the expectations of the T1D community. Once again, JDRF made clear to FDA that it was still falling short.

Shuren, who plans to leave the FDA by the end of 2024, remembers this period well. “JDRF asked me to meet with them and they laid out their concerns,” he recalled. “In particular, they talked about problems with the office and the FDA team. So, I started having a series of meetings with the team, JDRF, and others; and it became pretty clear to me, although they were well-intentioned, we were not engaging in a constructive manner.”

Shuren got the message, and he made his move. He decided to pull oversight of the entire AP project from the original group. “I determined that it made more sense to move the AP technology [and insulin pumps] to another part of the center that dealt with other diabetes related technology like glucose meters. So, I put all diabetes related technology in one place.”

Stayce Beck, a bioengineer, was the lead member of the new team. “The ads changed the trajectory of my life,” she said years later. “It spurred FDA to take the AP and give it to my group and they asked me to lead the project.” Early in 2012, the new team took the JDRF draft guidance “to heavily resource our version” and “to finalize it in a way that aligned with what they were thinking.” It took another 10 months but finally, in November 2012, FDA issued a new and improved guidance.

Finally, the AP community had leaders at the FDA who made clear they were there not to obstruct but to facilitate while staying true to the agency’s highest standard of ensuring any approved system was deemed safe and effective.

Said Shuren: “I do give JDRF credit for pushing, for saying there’s a real need for this, and raising the clarion bell on concerns with the center.”

“The companies would always tell us that FDA would never allow this to happen,” said Kowalski. “But now they had a guidance document from FDA saying it can happen if you do these trials this way, and the trials were the right ones, so the guidance was a seminal moment. It was huge.”

“The guidance opened the floodgates,” said Dutta. “Now companies knew how to do trials, who to recruit, how long to do them for, what do they need to prove. Now people knew there was a path as opposed to driving in the dark.”

Indeed, the research community did move into overdrive. Outpatient trials were launched and completed in waves. Kovatchev, Cobelli , Buckingham, Hovorka, a team at Mt. Sinai in New York and the Mayo Clinic in Minnesota, Ed Damiano at Boston University, all conducted trials of patients in different settings with different algorithms and system components between 2012 and 2016. They started with overnight outpatient trials, moved on to multiday in-home trials, and then to three-month and then six-month “real-life” trials. There were trials in diabetes camps and trials at ski camps. The pace was dizzying but the results were exhilarating. Time and again, the trials demonstrated both the safety and efficacy of automated insulin systems.

[10] Senators Susan Collins (R-ME) and Jeanne Shaheen (D-NH) chaired the Senate Diabetes Caucus and Representatives Diana DeGette (D-CO), and Ed Whitfield (R-KY) chaired the Congressional Diabetes Caucus.

Chapter 11: The Hackers

Progress was happening. But for many with T1D it was agonizingly slow. Most could only wait. Some decided they were tired of waiting.

Bryan Mazlish had founded an automated stock trading business where he and his team wrote algorithms to buy and sell stocks. His wife Sarah had been diagnosed with T1D when she was 12. In 2011, their middle child, Sam, was diagnosed with T1D at age 5.

“I am embarrassed to say how little I understood what my wife was going through on a daily basis until I was helping my son,” said Mazlish. “I never really understood the pernicious impact that the disease has on someone. You’re essentially driving a car on a curvy mountain road 24 hours a day, seven days a week with no time off even when you’re sleeping.”

Sam started wearing a CGM, but the couple was frustrated with the technology, largely because it was hard to hear the system’s alarm go off at night signaling a problem. “We wanted to be able to hear our son’s alarms at night and also monitor his glucose,” said Mazlish. Working with his father-in-law and brother-in-law, both engineers, they managed to figure out a way to remotely monitor Sam’s glucose levels and ensure they would hear the alarm when Sam was in trouble.

Excited about this step forward to make their son’s life safer, Mazlish asked his wife what else they could do, “and she said she had just had a rough night and if she could just wake up with a perfect blood sugar every morning that would be amazing.” The lightbulb went off.

Sarah and Bryan designed a do-it-yourself (DIY), or hacked, prototype artificial pancreas using an android phone and an algorithm they created to communicate with the pump and CGM. Sarah was the beta tester and wore the system the first night after the pair felt they had debugged it. “The first night she wore it, she slept one of the best nights of her life, but I was basically up all night monitoring the machine, making sure nothing bad happened,” said Bryan. Sarah used it for a few more months, they continued to tweak it, and when the couple were completely comfortable in it, they put it on Sam.

One other person was excited about the system. Aaron Kowalski. “Bryan calls me one day and says he wants to show me what he did. So, he comes to 26 Broadway (JDRF’s headquarters) and shows me his cell phone and it has his son’s glucose reading on the screen. I was like, ‘what the hell, what is this?’ And he says, ‘Oh, it was really easy. I just hacked it.”

Word started getting around about the system in T1D circles and Mazlish tried to get companies interested. “I talked to all the existing device makers, CGM companies, pump companies, and showed them what we had done, the data from Sam and Sarah and how amazing it was,” said Mazlish. “I said I would come work for them. But there were a lot of questions about whether FDA would approve it and a lot of hesitation and not a lot of urgency.”

Indeed, FDA was direct. “We were absolutely unenthusiastic,” said FDA’s Stayce Beck. “We did not want people to resort to these systems. We wanted to give them approved options.” If it was made available to others, even at no cost, the device would require FDA approval. In the end, the agency sent Mazlish a message: if you don’t try to promote this or make it widely available, we won’t act against you.

But by then others had taken up the DIY mantle.

Dana Lewis was diagnosed with T1D when she was a 14-year-old girl in high school in Huntsville, AL. As an adult she moved to Seattle in 2010 where she worked for a nonprofit hospital system helping how it and its doctors could effectively use social media as a communications tool. In 2013, she met Scott Leibrand, a network architect who would become her husband. Frustrated with existing diabetes management tools, they began to tinker with existing systems and in the summer of 2014, they met Ben West, another hacker who joined their work.

“By late November we had closed the loop,” said Lewis. They had in effect created a hacked AP system that predicted and prevented high and low blood sugars. In short order, the team built a website to tell the world about the OpenAPS, an open-source system that would allow others to replicate their creation. In this case, the FDA was powerless. “We weren’t distributing anything but ideas,” said Lewis. “That’s not against FDA rules and we felt pretty comfortable with open source and sharing it.” A few years later, a programmer, Pete Schwamb, took on the challenge of developing a hack that would allow users of the popular Insulet Omnipod pump to also adopt a hacked AP system.

Brewer, who would eventually leave JDRF to start Bigfoot Biomedical, with Mazlish and others believe the DIY movement played an important role in motivating the FDA to move faster.  “DIY drove FDA crazy,” said Brewer. “It caused the FDA to put pressure on companies to move faster, they wanted it to go away and so they became advocates for companies to step up their game and move quickly to get away from what the FDA thought was the ‘crazy’ stuff people were doing themselves.”

Chapter 12: The Patients

Starting in 2008, hundreds of kids and adults with T1D around the world volunteered for clinical trials to test the emerging AP systems. Early on, the risks were real, and it took a lot of trust for people to rely on new-fangled technology to keep them safe. But they did it anyway.

Tom Brobson was diagnosed with T1D in July 2004 at the age of 44. “I was not the typical adolescent with T1D, and I was just trying to cope and remap my life. Then I saw Mary Tyler Moore8 being interviewed by Larry King on CNN and I donated to JDRF.”  One year later, while working at Virginia Tech as a development official, he heard that JDRF was seeking to hire a major gifts officer, so he applied for the job, and was hired. Brobson would soon become one of the most visible and inspiring JDRF figures when he became one of the first persons in the world to wear a closed loop AP system as a participant in Boris Kovatchev’s hospital trial in 2008.

Tom Brobson and Boris Kovatchev in one of the early trials.

“They have two laptops set up,” he recalled of the experience. “I have two sensors on me. And I have an insulin pump. The two sensors were in case one failed. It was crazy just to walk down the hall to the bathroom because they had to put everything on big metal cart and wheel me down with all the wires connected. I ate dinner and watched the algorithm do a better job than I had done ten days prior. Most of all, that night, the big difference was I slept. When I ran things, I bounced low like six times through the night but when the system took over and ran things I avoided all that. And I remember sitting there texting Aaron and saying, ‘Oh my God, this thing works.’ It was a powerful moment.”

Brobson was also in the first UVA outpatient trial where the technology had transitioned to a portable system with a smart phone linked wirelessly to a glucose sensor and insulin pump. “We’re walking around with these phones with a stoplight design, a red, green, and yellow light with green indicating you were in a safe range, yellow that you were moving in the wrong direction, and red meaning that you were having a low or high blood sugar,” recalled Brobson.

Joshua Davis was diagnosed with T1D in 2009 when he was 11 months old. Joshua’s dad Brian had been diagnosed with T1D in 2004 right before his tenth high school reunion, so the family had some familiarity with the disease. But dealing with an infant was a whole different challenge.

“It was heartbreaking and just completely devastating,” said Shannon Davis, Joshua’s mom. “I went from learning how to carb count breast milk and baby food at diagnosis, because at the time that’s all he was eating, then carb count goldfish cookies when he was a toddler, figuring out how many goldfish did he throw on the floor. You see other moms pushing their babies in a stroller and they’re just kind of handing ’em food while you’ve got to stop and count each tiny goldfish that you’re giving your child. And then the sleep, we just didn’t sleep. Typically, we would check his blood sugar before he went to bed, and then we check it again at about 10 pm, then about 1 am, and then about 3 am, and then we would try to get until 7 am. Joshua actually learned how to eat a banana in his sleep to elevate his blood sugars.”

At age 5, Joshua had the opportunity to participate in a UVA outpatient trial. His parents said the choice was his. “I was really excited because I was already doing sports so the idea of being able to run around without having to worry about going low or having to stop and eat a snack in the middle of playing was a freeing feeling,” he recalled.

Joshua Davis sleeping through the night.

Joshua and Shannon headed to the Wintergreen Resort in Virginia where they would room together while Joshua wore the AP system. For several years Joshua had been on a CGM and a pump, so Shannon was used to responding to the alerts and helping her son make safe decisions on dosing insulin. But the camp was a big change.

“I can vividly remember the first night of the trial,” said Shannon. “The doctors said we needed to trust that they are watching our kids on a computer, and they told us we were not to respond the way we normally would so the system can do what it needs to. And I remember when Joshua woke up that first morning, he said he felt so good and I said, ‘Baby, you slept the whole night.’”

Alecia Wesner was diagnosed with T1D in 1979, and her parents became active with what was then called JDF. Growing up, she had two dangerous lows where she fell into unconsciousness. Eventually, she wound up in New York City in 1998. Soon, though, she was diagnosed with diabetic retinopathy, one of the common complications of T1D that can lead to blindness, especially if a person does not maintain excellent blood-sugar control and seek available treatments. As the AP Project took root, Alecia joined the local JDRF chapter. “At one of our Board meetings, Tom Brobson spoke, and he talked about being in trials at UVA, and I said I need to do that.”

She wrote to the UVA team, and they told her there was an upcoming trial that she qualified for at UVA. Months went by and then she was notified that they were no longer needed for that trial because it was no longer necessary as the technology had moved forward. But the email concluded by revealing that the next phase of AP trials were moving to other test locations including Mt. Sinai Hospital in Manhattan and she should contact the lead principal investigator at Mt. Sinai, Dr. Carol Levy. “When I got the email, I was sitting in her office.” Carl Levy was Alecia’s endocrinologist. Alecia got into the trial.

Alecia Wesner in her first clinical trial.

“It was amazing. I went to work every day. Nothing in my life was different. And then after work I would rush back to the hotel and they would set everything up and once we were in bed for the night, we got to turn the unit on. And they observed us while we slept. I knew from my own experience that I had been having problems going low at night then bouncing back up in the morning and sleeping through the alarms from my CGM. With this AP system, every day I woke up with glucose levels in range, so however my day started I was not chasing a high or low blood sugar.”

Looking back at her motives for volunteering, Wesner said: “Getting involved in AP trials to me was my chance to pay it forward for somebody else. I have lived 45 years with T1D, and I have lost friends to type one. And I think there’s an enormous responsibility that comes with being alive. I think there’s something comforting in knowing that my body was used for something that not only had the potential to make me healthier, but really was for other people. I do think there’s something to be said for doing good, feeling good, and this is what it felt like being part of trials.”

[8] Mary Tyler Moore had been diagnosed with T1D around 1970 at the age of 33. She soon became JDRF’s International Chairperson using her fame to raise awareness of the disease, increase private and federal research support, and testifying before Congress and inspiring others with the disease to live their lives fully and boldly

Chapter 13: The Finish Line

Spurred by the FDA guidance, goaded by the JDRF Animas deal9, the 800-pound gorilla finally come down from its mountain habitat and got into the game. Fittingly, it was FDA that played a key role in catalyzing the company’s involvement. “FDA would tell us flat out they were begging Medtronic, which was amazing, it was a 180-degree turn, like going from incredible headwinds at to having the wind at your back,” said Kowalski.

“In 2013 we reached out to Medtronic because they had a pump, a CGM, and an algorithm and we said we want to make this happen,” said Stayce Beck. “Let’s figure out what we need to do. So, we ended up meeting with Medtronic monthly to get them to accelerate. They originally weren’t planning on being in the market until 2019.”

In mid-2016, Medtronic announced that it had completed a trial of a hybrid closed loop system called the MiniMed 670G. The trial included 123 adolescents and adults at nine sites in the US and one site in Israel. It was a so-called pivotal trial, the last step in the drug and device development process before applying to the FDA for approval. Participants on the system for three months experienced a 44% reduction in time spent with a low blood sugar, a 40% reduction in dangerous low-blood-sugar territory, and an 11% decline with dangerous high blood sugars.

Just three months later, acting with unprecedented speed, on September 28, 2016, the FDA approved the 670G, and the Artificial pancreas Project had achieved its goal.

[9] In 2017, Animas exited the diabetes market without ever completing work on an AP. Speculation was that the company did not move quickly enough and was eclipsed by new iterations of the original Medtronic 670G.

Chapter 14: The Aftermath

Since the approval of the 670G, FDA has approved four additional systems, including the T:Slim X2 by Tandem, using the original algorithm developed by Boris Kovatchev, the Omnipod 5 by Insulet, using the algorithm developed years earlier by Frank Doyle’s team at UCSB, and Tandem’s Mobi, the Cam APS developed by Dr. Roman Hovorka’s team at the University of Cambridge in England, the iLet Bionic Pancreas that emerged from Ed Damiano’s team at BU10, and most recently, the Dean Kamen designed twiist from Sequel Med Tech. Thousands of others continue to use DIY systems.

As readers will recall, the landmark Diabetes Control and Complications Trial, or DCCT had found that people with T1D could reduce the risk of long-term complications by 35-76% through extremely tight blood-sugar control—keeping their levels in a range that largely eliminated dangerous highs and lows. But, as Brewer discovered when his son was diagnosed, meeting these ideal targets was nearly impossible with the technology in use in 2005.

Today, two decades after Brewer and Kowalski persuaded a skeptical JDRF to launch the AP Project, the latest available data leaves no doubt that their belief that developing AP systems would improve lives and reduce the risk of long-term health complications for people with T1D has been vindicated.

In June 2024, the T1D Exchange, a nonprofit dedicated to improving outcomes for people with T1D, completed a study11 of people living with T1D and found that children under 13, teens, and young adults between ages 18 and 29 using hybrid closed loop systems were achieving safe target blood-sugar levels 40-68% more often than those using insulin pumps and CGMs. Results were slightly lower for adults over 30 years old but still were statistically significant. The T1D Exchange data also showed that young people tended to have fewer severe low blood-sugar events on hybrid closed loops than those on pumps and CGMs.

“What we brought to bear is resulting in a safer and easier life for hundreds of thousands, and soon millions, of people with T1D, including my son, that is going to keep them safe until something like a cure comes along,” said Brewer.

[10] Work still continues on the elusive Goal Six multi-hormonal system.

[11] Automated Insulin delivery Use among 12,065 T1D Exchange Registry participants

Postscript

Pregnant women with T1D have a host of challenges and risks. They are eating for two, they have varying levels of stress, they have additional hormones impacting them systemically, including hormones that can upset blood-sugar control. In fact, one of two babies born to women with T1D have complications, most commonly preterm birth, large birth weight, and admission to neonatal ICUs. In 2008, Dr. Helen Murphy, a UK clinician who works with pregnant women with T1D, started collaborating with Roman Hovorka at the University of Cambridge, one of the early artificial pancreas research pioneers, who developed the CamAPS (artificial pancreas system) which was approved in the U.S. by FDA in May 2024.

Dr. Murphy’s focus was on using the AP systems to improve pregnancy outcomes for T1D Moms. Early systems were not sufficiently reliable to risk using with pregnant women, but by 2023 the time was ripe. Hovorka, Murphy, and colleagues around the UK conducted a trial of 124 pregnant women with T1D to test whether the Cambridge AP system could improve maternal glucose enough to benefit pregnancy outcomes. All users of this system experienced 10.5% more time in range throughout their pregnancies. Soon after, the UK’s National Institute of Health Care and Excellence issued clear guidance advising that all pregnant women in the UK be given access to pregnancy-specific AP systems. More than forty years after Bill Tamborlane asked Dean Kamen to develop a pump for his work with pregnant woman, the technology had become standard of care.

In February 2024, my daughter, diagnosed with T1D at age 14 in 2001, gave birth to her second child. She had her first in 2020. She wore the original DIY version of the Omnipod closed loop system in both pregnancies. Both babies were born healthy and on time.

“The loop system has truly transformed my life, making everyday activities easier and less stressful,” Emma said. “It allowed me to experience both of my “high-risk” pregnancies as “normal,” with tighter control over my blood-sugar levels so I could focus on enjoying the prenatal and postpartum journey. Day to day, this technology has given me greater stability and peace of mind, letting me live life fully without the constant worry of managing my diabetes.”