Hot from the Oven, ATTD 2026 Conference Catch-Up

Imagine teaching your thigh muscle to act like a pancreas. Right. I mean, it sounds like pure science fiction, right? You inject a sequence of genes into your leg. And suddenly the muscle tissue just learns how to sense your blood sugar and release insulin entirely on its own. It really does sound like something straight out of a futuristic novel. But as of a 7.45 AM presentation at a recent medical conference, this is the actual clinical reality we're heading toward. Wow. Well, welcome to a brand new deep dive, everyone. We have a stack of hot off the press presentations, detailing the absolute cutting edge of diabetes care. And this isn't just your standard textbook update, you know, with a few incremental tweaks to existing medications. No, not at all. Our mission today is to explore four massive leaves that are completely rewriting the roles of chronic disease management. We're looking at smarter, wearable hardware, artificial intelligence algorithms, making independent dosing decisions. Disease modifying drugs. And finally, that genuinely mind blowing gene therapy. And you know, if you're listening and you don't have diabetes, you might think this specific topic doesn't really apply to you. Right. Totally. But this is actually a blueprint for how your own future healthcare is going to look. You're getting a front row seat to watch modern medicine transition from episodic reactive care, where you only see a doctor after something breaks to proactive, continuous and even localized cellular therapies. So to understand how we get to reprogrammed biology, we really have to start with the hardware people are currently wearing. So Dr. Viral Shaw presented data on the Medtronic Mini Med 780 G. Yes, the automated insulin delivery system. Exactly. But the focus here isn't just on having a shiny new gadget, right? It's about how the physical hardware fundamentally alters human psychology. That's the key. Dr. Shaw compared real world data from two different continuous glucose sensors that communicate with this specific pump. OK. So the older model is the Guardian for which launched back in April 2023. And the brand new model is the instinct sensor which launched in December 2025. Oh, so very recent, very recent. They looked at 14,000 real world users in the US who transitioned to the new instinct sensor and the outcomes were, well, they were exceptionally strong. Users achieved a 75.6% time in range. OK, let's clarify that metric for a second for everyone time and range or TIR. That's the percentage of the day a person's blood sugar stays within a healthy target zone, right? Usually between 70 and 180 milligrams per decilator. Correct. And even more impressively, they saw a 51.8% time in tight range. Wow. Yeah, which narrows that goalpost even further. So between 70 and 140 overall 72% of the individuals using the instinct sensor hit the recommended clinical goal for glycemic control. I was looking at those outcomes before we started and I noticed something very specific about the comparison. The actual jump in time and range when people switch from the older sensor to the new one was from 75% to 77%. Right. I mean, a 2% bump is an improvement. Sure. But isolated like that, it didn't immediately scream medical revolution to me. Well, that 2% is deceptive if you only look at the final glycemic number. The real victory, Dr. Shaw highlighted was in user compliance, which ultimately stems from trust. Trust in the machine. Exactly. The instinct sensor was designed to be significantly more user friendly and reliable. And because of that, sensor use, the actual amount of time people kept the device physically attached to their bodies jumped from 90% with the old sensor to 98% with the new one. Oh, that makes sense. It's some it's kind of like upgrading from standard cruise control to a fully self driving car. That's a great way to look at it. Like the autonomous tech might be vastly superior, but the car only drives itself if you actually trust the system enough to take your hands off the steering wheel. Right. If the system constantly beeps at you or feels unreliable, you're just going to turn it off and go back to driving manually. That analogy captures the data perfectly because users trusted the sensor enough to wear it 98% of the time, the system spent an extra two hours a day in automated mode. Two whole hours. Yeah. And that additional time allowed the algorithm to deliver more automated insulin boluses, you know, those tiny micro correction doses of insulin. The system actually went from delivering 13 of those corrections a day to 17. So the machine was just doing the heavy lifting in the background. It caught the tiny blood sugar rises and fixed them before they ever became major spikes. Exactly. The underlying takeaway here is that the automated algorithm matters more than the raw accuracy of the sensor provided the sensor is comfortable and trustworthy enough that people will actually keep it on. I mean, hardware is useless if it just sits on a nightstand. It really is. And that concept of hardware driving behavior leads directly into Professor Emma Wilmot's presentation on the Freedom 2 trial. Yes. Because we just discussed automated pumps for type one diabetes where the body produces zero insulin. But what about type two diabetes where the body produces insulin, but has just become profoundly resistant to it. Right. So Professor Wilmot looked at adults with type two diabetes who are really struggling to meet their targets. And these patients were heavily medicated like a maximum therapies. Yeah, they were on modern combination therapies. So SGLT2 inhibitors, which helped the kidneys excrete excess glucose, GOP ones, which stimulate insulin and slow digestion, and basal insulin, which provides a steady background dose. They were basically throwing the kitchen sink at the disease biologically. Exactly. But behaviorally, they were still relying on traditional fingers, sick blood tests, which is so tough because the finger stick only gives you a static snapshot. It tells you where you are at that exact second, but it doesn't tell you the velocity or the direction your blood sugar is moving. Right. So the trial randomized these patients. One group kept doing finger sticks and the other was given a continuous glucose monitor, the freestyle, Libre 3 and the results were fast, right? Very fast. Over just four months, the group using the continuous monitor saw their HBA1C. So the three month average of their blood sugar drop from 8.8% to 8.1%. That's a big drop. Huge. Their time and range jumped from roughly 40% to 54% and crucially, they achieved this without increasing their risk of dangerous low blood sugars. The mechanism behind that improvement is the most fascinating part of the study to me because the medication doses between the two groups were virtually identical, right? Yes, they were. So the gadget literally changed their daily habits. I saw the accelerometer data show the continuous monitor users took an extra 12 minutes of light physical activity per day by that four month mark. And their dietary questionnaire showed demonstrably healthier food choices because they were suddenly equipped with a real time biofeedback loop. I mean, if you eat a bowl of pasta and watch your phone immediately display a steep vertical spike in your glucose, you learn a localized lesson about your own metabolism. It makes it real. It does. And conversely, if you take a 10 minute walk after a meal and literally watch the trend arrow point downward, you are immediately rewarded for that behavior. The hardware empowered them to make micro adjustments to their lifestyle because cause and effect were finally visible. But you know, staring at a biofeedback loop all day creates its own unique problem, cognitive overload. Exactly. If you're one of the millions of people who don't wear an automated pump, knowing your blood sugar is only half the battle, you still have to do the math to fix it, which is exhausting. It is people who rely on multiple daily injections are constantly calculating carbohydrate ratios, stress levels, physical activity, and residual insulin from their last shot. And the clinical term for this is decision fatigue. You are essentially asking a patient to manually operate as their own pancreas, right? As Professor Andrej Jenesh pointed out, this unrelenting cognitive burden inevitably leads to burnout and delayed intensification of treatment. People simply get tired of doing complex algebra before every single meal, which introduces the second massive leap in the data we're diving into today. Software that steps in as a co-pilot. Professor Jenesh presented the Modi dose study. This is an artificial intelligence algorithm that aggregates a participants continuous glucose data and delivers daily safety guards or weekly dose head recommendations directly through a smartphone app. And the crucial detail here, which kind of blew my mind, is that it does this entirely without physician involvement. Right. The AI independently analyzes the trends and instructs the patient on exactly how much insulin to inject. Wait, hold on. Let me just pause right here. You were saying an AI is making daily dosing decisions for a hormone that is highly lethal, if overdosed and no human doctor is signing off on these adjustments. That's exactly what I'm saying. Isn't that a massive paradigm shift? Like how did that even get past a safety review board? How is that safe? Well, that inherent skepticism is exactly why this 10 week feasibility study was required in the first place. They tested the Modi dose algorithm on both type one and type two patients using daily injections and the primary endpoint was safety. Specifically, they needed to prove the AI could reduce severe hypoglycemia. Those are the dangerous events where blood sugar drops below 54 milligrams per deciliter, effectively starving the brain, a fuel. And what did the safety data actually show? Did it work? It showed a significant reduction in severe hypoglycemia alongside dramatic improvements in overall glycemic control. The HBA1C dropped by almost 0.4% across the board. Wow. And for patients who entered the trial with an HBA1C over 8% the drop was nearly 0.9%. The AI optimized their insulin safely by specifically avoiding a common clinical pitfall called over basalization. Oh, over basalization. I've heard of that. That happens when patients blood sugars are running high overall. And the temptation is to just crank up their long acting background insulin to force the numbers down right? Exactly. but that blanket approach usually leads to crashing low blood sugars in the middle of the night. Precisely. It's a lazy fix. The AI, however, was capable of making surgical, targeted adjustments instead. It analyzed the exact times of data patient was struggling and adjusted the short-acting meal time doses rather than just flooding the system with background insulin. The psychological impact of having that AI copilot is super clear in the patient satisfaction data, too. One hundred percent of the participants requested to keep the algorithm after the study ended. Every single one. Every single one. They reported feeling safe and they felt the crushing burden of managing the disease was finally lifted. The expertise of an endocrinologist was localized into an app in their pocket. And we are seeing a structural shift here from episodic care, you know, waiting three months to see a specialist to review your data and adjust your dose to continuous daily data-driven optimization. But while software and AI are undeniably incredible at optimizing the math of how much insulin you need, there is a ceiling to how much software can do if the underlying biological system is still broken. Right. Which brings us to our third leap. Exactly. We are transitioning from managing the math with software to actively resetting the metabolism with disease modifying pharmaceuticals. So let's talk about professors to Phonodelle Prado's presentation on the surpass early study. Yes. We appreciate the data from this study. We first need to establish a concept known in endocrinology as the legacy effect. Right. And this goes back to a landmark trial called the UKPDS. The core finding there was that getting a patient's blood sugar under tight control immediately after a type 2 diabetes diagnosis provides a biological legacy of protection. Yes. Even decades later, those early control patients suffer far fewer microvascular and macrovascular complications. The eyes, kidneys and hearts are protected. The biological foundation of the disease is established early. Let's put that in practical terms for everyone listening. So the legacy effect is basically like if you discover a leak in your roof, right, you have two choices. You can fix the hole immediately or you can just put a bucket under it and diligently empty that bucket for five years before you finally patch the roof. That's a good way to frame it. Even if you never let the bucket overflow, the prolonged moisture still rots the floorboards and compromises the foundation. The early intervention prevents the structural damage. And Sir Pass early aimed to be that immediate structural repair. They utilized a medication called terozepatide in adults who had been diagnosed with type 2 diabetes within the last four years. Okay. Perzepatide is a dual GIP and GLP1 receptor agonist. Let's break down that mechanism a bit. What does a dual agonist actually do inside the body? Well, instead of just targeting one hormonal pathway to stimulate insulin and lower blood sugar, it knocks on the blood sugar. And two different cellular doors simultaneously. The GIP and GLP1 hormones work synergistically. Okay. So a double approach. Right. Think of it like rebooting a frozen computer system and clearing the cash at the same time. It allows the pancreas to breathe and function without fighting through immense metabolic resistance. And they compared this early use of terozepatide against intensified conventional care. The conventional care group received standard approved diabetes medications and over two years they did very well. They hit an average HBA1C of 6.3% and lost about 6.5 kilograms. Historically, any doctor would view that as a massive clinical success. Oh, absolutely. By traditional metrics, yes, that is excellent management of the disease. But the terozepatide group completely redefines what success looks like. Their HBA1C plummeted by two full percentage points to land at 5.56%. Incredible numbers. I mean, that is not a well-managed diabetic number. That is a strictly non-diabetic level. And their weight loss averaged 16 kilograms. Nearly 70% of the people on terozepatide achieved a completely normal A1C compared to only 26% in the conventional care group. And Professor Doprado emphasized that this goes far beyond just driving blood sugar down. It is a holistic metabolic reset. The patients on terozepatide saw profound improvements in their lipid profiles, their HDL, VLDL and triglycerides normalized. Their blood pressure improved. And honestly, the most critical metric was the significant improvement in beta cell function. Oh, wow. So the cells themselves? Yes, the actual cells in the pancreas responsible for producing insulin were fundamentally healthier. The drug didn't just mask the symptoms. It rewound the metabolic damage at a cellular level. That is just amazing. But while terozepatide is offering a complete reset for type 2 diabetes, we really have to acknowledge the biological reality of type 1 diabetes. It is a totally different beast. It is. Type 1 is an autoimmune condition. The body's own immune system has actively hunted down and destroyed the insulin producing beta cells in the pancreas. Which brings us to Dr. Jeremy Pettis, who presented the final study of our deep dive today. And he laid out a pre-sobring reality. Even if a type 1 patient has the absolute perfect automated AI pump we discussed earlier, they still face massive physiological hurdles. Type 1 patients still have an elevated risk of cardiovascular disease. Two thirds of them are overweight or obese. And as the years go on, insulin resistance becomes a universal reality. Managing the disease from the outside with pumps and exogenous insulin is just no longer enough. We need it totally outside the box approach. And that work around that outside the box approach is gene therapy. Specifically, the progress study developed by a research group called CUEA. Okay, so when I first saw gene therapy in the source material, my immediate assumption was CRISPR. A lot of people assume that. Yeah, like are scientists permanently editing someone's DNA? And even if they do, wouldn't a type 1 diabetic's immune system just destroy the new insulin producing cells anyway since it's an autoimmune disease? Well, Dr. Pettis addressed those exact hurdles perfectly based on their data. First, this is fundamentally not CRISPR. This is not gene editing where you slice into the host's genetic code. This is gene addition. They use a non-pathogenic virus, an adenine associate virus, or AAV. As a microscopic delivery vehicle. And it carries two specific genes, one for insulin and one for an enzyme called glucokinus. And they inject this delivery vehicle directly into the quadriceps, right? The large muscle tissue in the leg. Yes. The virus carries the genetic payload into the nucleus of the muscle cell, but it does not integrate into the patient's own DNA. It sits there as an episome. An episome. Right. It's essentially a standalone ring of DNA that resides next to the host genome, quietly producing proteins without risking a catastrophic mutation in the patient's actual genetic code. It's like a guest bringing their own chair to the dinner table instead of tearing down the whole dining room to build a new one. Exactly. But that still leaves the immune system problem. Historically, anything resembling a cellular transplant required chronic, lifelong, immunosuppressant drugs. Well, the progress study bypasses that by using a short course of an immune modulator. Patients receive a drug like Toselizumab, which is an IL-6 inhibitor, for just two to three months. Oh, just a few months. Just a few months. This temporarily calms the aggressive immune response just long enough for the viral delivery to settle in and establish that episome. Once the episome is stable in the muscle tissue, the immune drugs are stopped entirely. The biological mechanism of how those two genes function together is just staggering. The glucocanase gene acts as a biological glucose sensor right there in the thigh muscle. Right. They switch on when blood sugar rises, prompting the muscle to secrete the insulin from the second gene. And then when the blood sugar drops back to a normal level, the glucocanase sensor switches off, halting the insulin. They are essentially teaching the leg muscle to act as a functional, autonomous, pancreas. And the preclinical data supporting this mechanism is incredibly robust. They tested this specific intramuscular injection on diabetic dogs. Helping a single procedure, those dogs maintain completely normal blood sugars for up to seven years. Wait, really? Seven years. Seven years of physiological independence from a single treatment. And the timeline for humans. Clinical trials are actually slated to begin this year in 2026. Wow. Initially, they're enrolling type 1 patients who are currently using advanced automated pump systems. This allows researchers to ensure safety while gathering precise real-time data on exactly how much the exogenous insulin requirements drop once the muscle tissue starts producing its own. Unbelievable. We have covered an immense amount of ground today. We started by looking at how improving the user experience of a continuous glucose sensor builds trust, allowing patients to stay in automated delivery modes for extra hours a day. Yes. And we explored how continuous biofeedback in the freedom to trial naturally incentivize people to make healthier lifestyle choices. We examined how artificial intelligence can step in and alleviate the crushing cognitive burden of daily injections, making precise dosing decisions while actively preventing overbasalization. We also looked at the surpass early data, proving that dual-iginous medications can reboot the metabolism of type 2 patients early on, leading to non-diabetic blood sugar levels and profound weight loss. And finally, we unpacked the mechanics of using a virus to literally teach your leg muscles to produce insulin. So a lot to take in. It is. And if you're listening, remember that none of this is isolated medical trivia. This is a blueprint for the future of chronic disease management. The era of episodic visits and retroactive symptom management is ending. We are moving toward continuous, personalized, and cellular level care. Absolutely. And before we sign off, consider the broader implication of that final gene therapy study. For the last century, medical science has been obsessed with building external machines and synthesizing chemicals to perfectly mimic the human body. But if gene therapy like the progress study succeeds, the ultimate medical device the future won't be something you wear on your belt or a pill you swallow. Right. It will simply be your own reprogrammed biology. If we can teach a muscle to be a pancreas, what other inherent limitations of the human body are just waiting to be rewritten? That is something to definitely mull over. Thank you for joining us on this deep dive everyone. Keep questioning the boundaries of what's possible, and we'll see you next time.