Dexcom’s Type 2 Report: The Survey, and the Trial Behind It
Dexcom’s Type 2 Report: The Survey, and the Trial Behind It
A new survey of 2,502 people with Type 2 diabetes finds most know little about CGM. The stronger evidence on whether CGM helps them comes from a randomized trial the report cites.
Type 2 and CGM: the survey and the trial
Dexcom surveyed 2,502 adults with Type 2 diabetes and 888 clinicians in eight countries. It is a company survey of what people say, not evidence that CGM works. Disclosure: Dexcom sponsors the Juicebox Podcast.
What it shows
- Survey (Dexcom / HarrisX, April 2026): 55% of people with Type 2 said they have limited knowledge of CGM; 72% did not know GLP-1s and CGM can be used together.
- Survey: 90% of U.S. clinicians cited coverage or reimbursement as a barrier, versus 53% in the other seven countries.
- CONNECT randomized trial (283 adults with Type 2 not on insulin, 26 weeks): A1C fell 1.6 points with a Dexcom G7 versus 0.7 with routine care, a 0.9-point difference.
- JAMA Network Open cohort (9,258 adults): wearing CGM more than 270 days a year was associated with a 1.52-point A1C drop over 12 months, versus 0.63 without CGM.
What it does not show
- CONNECT was sponsored by Dexcom and, as of this writing, is a conference abstract, not a full peer-reviewed paper.
- Six months of data. Nothing yet on whether the A1C drop lasts or changes complications.
- The cohort shows an association, not cause, and three of its seven authors work for Roche Diagnostics.
What it means for this community
- The report is about Type 2, but many people with Type 1 have family and friends living with it.
- The survey’s clearest signal is how many people with Type 2 say they know little about CGM.
- Whether a CGM fits someone’s treatment, and whether insurance covers it, is a question for their doctor.
Sources: Dexcom State of Type 2 Report 2026 (Dexcom data on file) · Oser T, et al., CONNECT, Diabetes 2026;75(Suppl 1):1170-OR · Hirsch IB, et al., JAMA Netw Open 2025 · ADA, Standards of Care 2026, Section 7
Educational only. Not medical advice. Talk with your doctor before making changes to your care. juiceboxpodcast.com · Full disclaimer
Dexcom released its 2026 “State of Type 2” report on September 28 at the European Association for the Study of Diabetes meeting in Milan. It is a survey. In April 2026, Dexcom partnered with the research firm HarrisX to poll 2,502 adults with Type 2 diabetes and 888 clinicians across eight countries, including the United States.
The headline finding is a knowledge gap. Only 33% of people with Type 2 said they know what a CGM is and how it could help. 28% said they do not know what one is at all. And 72% did not know that GLP-1 medications and CGM can be used together.
Disclosure: Dexcom sponsors the Juicebox Podcast. That is a reason to read this post, and the report, with care. A company survey measures what people say. It is not evidence that CGM works. For that, the report points to other research, covered below.
What the survey measured
Among GLP-1 users in the survey, 58% said they pair the medication with a CGM. Among GLP-1 users who are not on insulin, the report puts that figure at 27%.
Clinicians named cost. In the U.S., 90% cited coverage or reimbursement as a barrier to CGM, compared with 53% across the other seven countries. About one in four U.S. respondents with Type 2 said they had trouble paying for diabetes medication or supplies in the past year.
The roughly 300 current CGM users in the survey were positive: 94% said they manage their diabetes more independently, and 91% said they use the data to make treatment decisions. Those are self-reports from people who chose a CGM and kept wearing it. The survey says nothing about people who tried one and stopped.
What the trials show
The stronger evidence is the CONNECT trial, which the report cites. CONNECT randomly assigned 283 adults with Type 2 who do not use insulin, at 22 U.S. primary care practices, to either a Dexcom G7 or routine care with a fingerstick meter for 26 weeks. Dexcom sponsored the trial.
Participants started at an average A1C of 8.8%. A1C fell 1.6 points in the CGM group and 0.7 points with routine care, a 0.9-point difference. Time in range (70–180 mg/dL) was 62% with CGM and 41% without. The results were presented as a conference abstract at the American Diabetes Association’s Scientific Sessions in June 2026. A full peer-reviewed paper could not be found as of this writing.
A separate study in JAMA Network Open looked back at insurance and medical records for 9,258 adults with Type 2. People who wore a CGM more than 270 days a year saw A1C fall 1.52 points over 12 months, compared with 0.63 points in matched people who did not use one. That is an association, not proof of cause. Three of the study’s seven authors work for Roche Diagnostics, a diabetes device maker.
What is still open
CONNECT answers a narrow question: over six months in primary care, does adding a CGM lower A1C for people with Type 2 who are not on insulin? It does not yet show whether the drop holds past six months. A six-month extension is underway, according to the ADA. It also says nothing yet about complications. And participants knew which group they were in, which is unavoidable when the treatment is a device.
The ADA’s 2026 Standards of Care already recommend CGM for people with diabetes on “any diabetes treatment where CGM helps in management.” That part of the recommendation is graded C, below the A given for people on insulin. Randomized evidence like CONNECT is what could move that grade. Whether it does is up to the ADA.
How the survey was run
HarrisX surveyed 2,502 adults with Type 2 and 888 clinicians in April 2026, roughly 310 patients and 110 clinicians per country: Germany, Italy, Japan, Poland, Saudi Arabia, Sweden, the United Kingdom, and the United States. Clinicians were split about evenly between doctors and nurses.
The report does not say how respondents were recruited. Its unreferenced figures are cited as “Dexcom Data on File, 2026,” meaning the underlying data has not been published for outside review.
What this means for this community
Most of the Juicebox audience lives with Type 1 or cares for someone who does, and CGM is common there. This report is about the much larger Type 2 population, and plenty of listeners have a parent, partner, or friend in it. The survey’s clearest signal is how many people with Type 2 say they know little about CGM.
The trial evidence for CGM in Type 2 without insulin is newer and still maturing. For anyone with Type 2 who is curious, the questions for their doctor are whether a CGM fits their treatment and whether their insurance covers it. This post is for education only. Talk with your doctor before making any changes to your care.
Sources. The Dexcom report and press release were used for the survey figures only. Clinical claims are cited to the underlying research.
Survey. Dexcom. State of Type 2 Report: Global Access and Attitudes to Diabetes Technology, 2026. Conducted with HarrisX, April 2026; Dexcom Data on File. Report page · Press release, September 28, 2026.
Randomized trial. Oser T, et al. CGM for Adults with Type 2 Diabetes Not on Insulin Therapy: The CONNECT Randomized Controlled Trial. Diabetes 2026;75(Suppl 1):1170-OR, conference abstract. Sponsored by Dexcom. Additional figures: American Diabetes Association; Cleveland Clinic Journal of Medicine.
Records study. Hirsch IB, Garg SK, Repetto E, et al. Continuous Glucose Monitoring Frequency and Glycemic Control in People With Type 2 Diabetes. JAMA Netw Open 2025;8(10):e2539278. Retrospective, propensity-matched. Three authors affiliated with Roche Diagnostics.
Guideline. American Diabetes Association. 7. Diabetes Technology: Standards of Care in Diabetes—2026. Diabetes Care 2026;49(Suppl 1), recommendation 7.15.
How to read this evidence. A randomized trial is the strongest design here, but one sponsored trial reported as an abstract is an early result. The records study shows an association, not cause: people who keep wearing a CGM may differ in other ways from people who do not. The survey reports opinions and self-described behavior only.
This post is for educational purposes only and is not medical advice. Nothing here is a recommendation to start, stop, or change any device or medication.
Listen to the Juicebox Podcast
Conversations about diabetes, five days a week, since 2015.
Read the full disclaimer
FDA Approves Kerendia for Kidney Disease in Type 1 Diabetes
FDA Approves Kerendia for Kidney Disease in Type 1 Diabetes
Bayer’s finerenone is now approved for adults with chronic kidney disease tied to type 1 diabetes. The approval rests on one six-month trial that measured a lab marker, not kidney failure.
On September 16, 2026, the FDA approved Kerendia (finerenone), a once-daily pill from Bayer, for adults with chronic kidney disease associated with type 1 diabetes. Bayer calls it the first new treatment for this group in more than 30 years.
The approval rests on one randomized trial, FINE-ONE, published in the New England Journal of Medicine in March 2026. That trial did not count kidney failure. It measured albumin in the urine, a lab marker of kidney damage. The label, as quoted by Bayer, says the drug reduces that marker, which is expected to lower the risk of kidney decline and end-stage kidney disease.
The drug itself isn’t new. According to Bayer, it has been approved since 2021 for kidney disease in type 2 diabetes, and since 2025 for a type of heart failure. What changed is who it’s approved for.
What FINE-ONE found
FINE-ONE enrolled 242 adults with type 1 diabetes, reduced kidney function (eGFR 25 to under 90), and elevated urine albumin (a urine albumin-to-creatinine ratio, or UACR, of 200 to under 5,000 mg/g). Everyone was already taking an ACE inhibitor or an ARB. Participants were randomized to finerenone, 10 or 20 mg a day depending on kidney function, or placebo.
Worth naming plainly: Bayer funded the trial and makes the drug, and several of the paper’s authors are Bayer employees.
The result: over six months, urine albumin fell 25% more with finerenone than with placebo (95% CI for the ratio, 0.65 to 0.87). Bayer’s release quotes the same comparison at single time points instead, 22% at month 3 and 28% at month 6. The trial tracked three things worth knowing. Tap each.
What nobody knows yet
FINE-ONE shows finerenone lowers albumin in people with type 1 diabetes. On its own, it does not show fewer cases of kidney failure. That link is borrowed from the type 2 trials. There are two fair ways to read that.
VIEW 1 The bridge is reasonable
In type 2 diabetes, FIDELIO-DKD (5,734 people, also Bayer-funded) found fewer combined kidney events with finerenone over a median 2.6 years: 17.8% versus 21.1% on placebo. The FINE-ONE design paper says regulators agreed urine albumin could bridge that evidence to type 1, as long as set criteria were met.
VIEW 2 A stand-in is still a stand-in
Six months and 242 people is too short and too small to count kidney failure, dialysis, or deaths. What is known about safety in type 1 also comes from those six months. No long-term outcome trial of finerenone in type 1 diabetes has reported.
Both can be true at once. The approval follows a logic regulators agreed to in advance, and the long-term answer in type 1 is still open.
The approval rests on a lab value moving in the right direction. That is a reasonable bet, and it is still a bet.
An editorial observationWhat this means for people with type 1
Kidney disease is a common complication. Bayer’s release puts it at roughly 20 to 30% of people in the U.S. with type 1 diabetes, and the FINE-ONE design paper says up to 40% develop it despite guideline-recommended treatment.
Kerendia is not an insulin and is not a blood sugar medication. The approval covers adults who already have kidney disease tied to type 1. The trial tested it on top of an ACE inhibitor or ARB, not in place of one, and it enrolled adults only.
Dr. Janet McGill of Washington University School of Medicine, the final author on the FINE-ONE paper, said in Bayer’s release that this group has had “limited options to address the risk of kidney disease progression.”
For anyone following their kidney health, the two lab numbers in this story are urine albumin (UACR) and eGFR. Whether any medication makes sense for a specific person, and how potassium would be watched, is a conversation to have with a doctor.
FINE-ONE results. Heerspink HJL, Birkenfeld AL, Cherney DZI, et al. Finerenone in Type 1 Diabetes and Chronic Kidney Disease. N Engl J Med 394(10):947–957, March 2026. Randomized, double-blind, placebo-controlled phase 3 trial, n=242. Funded by Bayer. NCT05901831.
FINE-ONE design. Heerspink HJL, et al. Rationale and design of a randomised phase III registration trial investigating finerenone in participants with type 1 diabetes and chronic kidney disease. Diabetes Res Clin Pract 204:110908, October 2023. Several authors are Bayer employees.
Type 2 outcomes. Bakris GL, Agarwal R, Anker SD, et al. Effect of Finerenone on Chronic Kidney Disease Outcomes in Type 2 Diabetes (FIDELIO-DKD). N Engl J Med 383:2219–2229, 2020. Randomized trial, n=5,734. Funded by Bayer.
1993 captopril trial. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD. The effect of angiotensin-converting-enzyme inhibition on diabetic nephropathy. N Engl J Med 329:1456–1462, 1993. Randomized trial, n=409.
Press release. Bayer, via Business Wire, September 16, 2026. Company statement, not peer reviewed.
How to read this evidence. FINE-ONE is a randomized trial, the strongest study design, but it is one trial, industry-funded, six months long, and its main endpoint is a lab marker rather than kidney failure. The kidney-outcome evidence for finerenone comes from people with type 2 diabetes. The approval covers adults with chronic kidney disease associated with type 1 diabetes; it does not cover children, and finerenone is not a treatment for blood sugar.
This post is for educational purposes only and is not medical advice. Nothing here is a recommendation to start, stop, or change any medication. Talk with your doctor before making any changes to your care.
Kerendia approved for kidney disease in type 1 diabetes
On September 16, 2026, the FDA approved Bayer’s finerenone (Kerendia) for adults with chronic kidney disease associated with type 1 diabetes. It is not a blood sugar drug, and the approval rests on a lab marker, not on kidney-failure outcomes.
What the research shows
- FINE-ONE (NEJM, 2026; 242 adults; Bayer-funded): urine albumin fell 25% more with finerenone than with placebo over six months.
- Everyone in FINE-ONE was already on an ACE inhibitor or ARB; finerenone was added on top.
- High potassium was the most common adverse event in FINE-ONE: 10.1% on finerenone vs. 3.3% on placebo; 1.7% stopped the drug for it.
- In type 2 diabetes, FIDELIO-DKD (NEJM, 2020; 5,734 people; Bayer-funded) found fewer combined kidney events with finerenone: 17.8% vs. 21.1% on placebo.
What it does not show
- Six months and 242 people: too short and too small to count kidney failure in type 1.
- The approval treats urine albumin as a stand-in for kidney outcomes, a link borrowed from the type 2 trials.
- In FINE-ONE, eGFR dropped more with finerenone at six months (a difference of 2.9 mL/min/1.73 m²); the paper reports values approached baseline after the drug was stopped.
For people with type 1
- The approval covers adults with kidney disease tied to type 1. FINE-ONE enrolled adults only, and the drug does not treat blood sugar.
- Bayer’s release estimates 20 to 30% of people in the U.S. with type 1 have chronic kidney disease.
- The two lab numbers in this story are urine albumin (UACR) and eGFR.
- Whether any medication fits a specific person is a conversation to have with a doctor.
Sources: Heerspink et al., N Engl J Med 2026, doi:10.1056/NEJMoa2512854 · Heerspink et al., Diabetes Res Clin Pract 2023, doi:10.1016/j.diabres.2023.110908 · Bakris et al., N Engl J Med 2020, doi:10.1056/NEJMoa2025845 · Bayer press release, Sept. 16, 2026.
Educational only. Not medical advice. · juiceboxpodcast.com
Listen to the Juicebox Podcast
Conversations about diabetes, five days a week, since 2015.
Read the full disclaimer
An AI-designed drug just reached Phase 3
An AI-designed drug just reached Phase 3
It treats a lung disease, not diabetes. It is still the clearest look yet at what artificial intelligence has changed about how new drugs get made — and what it hasn’t.
On September 9, 2026, the first patient was dosed in a Phase 3 trial of a drug called rentosertib. The trial, GENESIS-IPF-3, is planned to enroll 320 people with idiopathic pulmonary fibrosis across 47 centers in China, dosed once daily for 52 weeks, with the annual rate of lung-function decline as its primary endpoint. What makes it unusual is where the drug came from. AI systems were used both to identify the target it hits and to generate and optimize the molecule itself.
Insilico Medicine used one AI platform to rank drug targets and land on TNIK, a protein that had not been taken into clinical testing for lung fibrosis, then used a generative chemistry platform to design a compound against it. Humans still selected, synthesized, tested and optimized the molecules that followed. When the company published its Phase 2a results in Nature Medicine in June 2025, it noted that no AI-discovered drug had yet made it through a Phase 3 trial. This program is now in one.
There is no diabetes anywhere in this story. It is worth following anyway. “AI is going to speed up drug discovery” has been said to people with chronic conditions for years, and this is among the first programs far enough along to check the claim against something real.
What the Phase 2a trial showed
The Phase 2a trial, published in Nature Medicine in June 2025, randomized 71 adults with idiopathic pulmonary fibrosis to one of three rentosertib doses or placebo for 12 weeks. Insilico Medicine, which developed the drug, sponsored the trial, and company employees were among the researchers who designed the study and analyzed and interpreted the data. That is worth knowing before reading any of the numbers.
The primary endpoint was safety, not benefit. Lung function was a secondary measure: the 60 mg once-daily group gained a mean of 98.4 mL of forced vital capacity over 12 weeks while the placebo group lost 20.3 mL. The uncertainty was wide: the 95% confidence interval around the high-dose group’s mean change runs from 10.9 to 185.9 mL. That is a within-group estimate, not a confidence interval for the treatment effect against placebo, and this small safety-focused trial was never designed to establish efficacy. Sixteen of the 71 patients stopped treatment before week 12. Liver injury or dysfunction accounted for seven of those sixteen, and four of those seven were also taking nintedanib, an existing antifibrotic. The authors list the limits themselves: small arms, all participants in China, 12 weeks. One early trial does not settle whether AI changed anything. The pattern across the field is more informative.
The argument underneath all of this
The real question is not whether the software works. It plainly produces molecules, and those molecules keep clearing Phase 1. The question is which part of drug development was the bottleneck to begin with.
Two readings of the same evidence, both held by serious people:
THEORY 1 The slow part was chemistry
Finding and optimizing a molecule used to take years of synthesis and screening. Insilico reports reaching a preclinical candidate in 18 months. If chemistry was the real constraint, compressing it pulls in every timeline downstream, and the Phase 1 rates across AI-native companies are the first sign of it.
THEORY 2 The slow part was biology
Picking the right target has always been the hard part, and no platform has yet published evidence that it does that better than people do. The Phase 2 rates reported so far are in line with industry norms, and the 2026 Nature Reviews Drug Discovery Perspective finds no established clinical impact yet. On this reading, AI has sped up a step that was never the reason drugs fail.
Nobody can settle that yet, and the more careful reviewers say so directly. The number of AI-discovered drugs that have reached mid-stage trials is small enough that a couple of readouts would move the percentages noticeably. The 2026 Nature Reviews Drug Discovery Perspective calls this an absence of evidence rather than evidence of absence — not proof that AI fails to help, just no demonstration yet that it does, where it counts. Rentosertib’s Phase 3 runs 52 weeks before anyone sees a result.
Software has gotten much faster at proposing molecules. Nothing yet shows it is better at knowing which biology matters — and that is the step that usually decides whether a drug helps anyone.
An editorial read of the evidenceWhat this means if you live with type 1
Practically, nothing changes today. Rentosertib is being tested in a lung disease, is not approved anywhere, and has no application to diabetes. The useful part is calibration. When a headline says AI designed a drug, what has been demonstrated so far is a faster front end — target to candidate to first-in-human — not a faster or surer path through the trials that determine whether it helps.
That distinction lands harder in type 1 than in most conditions. Speeding up molecule discovery addresses one part of the problem, and the Diabetologia review lists the others: long and expensive trials, difficulty predicting who will progress, responses that vary widely between people, imperfect disease models, and unsettled endpoints. The same review notes that teplizumab, the first therapy approved for delaying type 1 diabetes, took about 30 years of testing to reach approval, including a Phase 2 trial that ran close to a decade. Cutting discovery from years to months does not shorten a trial that has to run long enough to see whether beta cells hold. It is reasonable to find this encouraging and still expect the same waiting. And if a product, test, or supplement is being sold to you on the strength of being AI-designed, that phrase describes how it was made, not what it has been shown to do. Talk with your doctor before making any changes to your care.
AI-discovered drugs in the clinic. Kp Jayatunga M, Ayers M, Bruens L, Jayanth D, Meier C. How successful are AI-discovered drugs in clinical trials? A first analysis and emerging lessons Drug Discovery Today, 2024. Authors are consultants at Boston Consulting Group. Retrieved via PubMed.
Discovery and Phase 1. Ren F, et al. A small-molecule TNIK inhibitor targets fibrosis in preclinical and clinical models Nature Biotechnology, published online March 2024. Insilico Medicine authors.
Aging-clock analysis. Zhavoronkov A, Galkin F, et al. Integration of proteomic aging clocks in a phase 2a clinical trial supports simultaneous geroprotective assessment Nature Biotechnology, September 2026. Insilico Medicine authors; exploratory analysis of 42 consenting trial participants.
Type 1 diabetes. Retrieved via PubMed. Shapiro MR, Tallon EM, Brown ME, Posgai AL, Clements MA, Brusko TM. Leveraging artificial intelligence and machine learning to accelerate discovery of disease-modifying therapies in type 1 diabetes Diabetologia, 2025;68(3):477-494 (published online December 2024).
Phase 3 first dosing. Insilico Medicine, Insilico Medicine doses first patient in GENESIS-IPF-3, company announcement, September 9, 2026 (NCT07687459 / CTR20262475). Trial size, center count and duration are as stated by the company.
The wider field, 2026. Bender A, Thomas MC, Scannell JW, et al. Artificial intelligence in drug discovery — what it is, where we stand and the path forward Nature Reviews Drug Discovery, August 2026. Peer-reviewed Perspective; academic and industry authors.
How to read this evidence. The Phase 2a trial is randomized and placebo-controlled, the strongest design here, but it was powered for safety rather than benefit, ran 12 weeks, and enrolled 71 people at sites in one country. The pipeline analysis is a survey of company pipelines rather than a controlled comparison, and its authors flag the small sample behind the Phase 2 figure; the 2026 Nature Reviews Drug Discovery Perspective is a peer-reviewed critique of that same evidence base. The aging-clock analysis is exploratory; its authors state plainly that they cannot separate aging effects from the drug’s anti-fibrotic activity within a fibrosis cohort. Rentosertib is an investigational drug. It is not FDA approved, not approved for any use in the United States, and has no indication in diabetes.
This post is for educational purposes only and is not medical advice. Nothing here is a recommendation to start, stop, or change any medication. Talk with your doctor before making any changes to your care.
An AI-designed drug reached Phase 3
The first patient was dosed in a Phase 3 trial of rentosertib on September 9, 2026. AI systems were used to pick its target and to design the molecule. It is being tested in a lung disease, not diabetes, and it is not approved anywhere.
What the research shows
- A randomized trial of 71 patients over 12 weeks, published in Nature Medicine in 2025, was designed to test safety. Side-effect rates were similar across the drug and placebo groups.
- Lung function was a secondary measure in that trial: the highest-dose group gained an average 98.4 mL while placebo lost 20.3 mL. The interval around that gain (10.9 to 185.9 mL) is a within-group estimate, not a comparison against placebo.
- A 2024 review in Drug Discovery Today found AI-discovered molecules clear Phase 1 at 80-90%, above industry norms, but succeed in Phase 2 at about 40%, the same as everyone else.
- Insilico Medicine, which developed the drug, sponsored the trial, and company employees were among those who designed and analyzed it.
What it does not show
- Anything about diabetes. Rentosertib is investigational, is not FDA approved, and has no diabetes indication.
- That AI picks better targets. Phase 2 is where drugs usually fail, and a 2026 Nature Reviews Drug Discovery Perspective finds no established clinical impact from AI yet.
- Long-term safety. In the trial, 16 of the 71 patients stopped early; liver injury or dysfunction accounted for seven of them.
If you live with type 1
- What AI has sped up is the front end: choosing a target and designing a molecule. The company reports 18 months from starting target discovery to naming a candidate.
- Faster discovery fixes one bottleneck. A 2025 Diabetologia review lists the others: long trials, hard-to-predict progression, varied responses, imperfect models, unsettled endpoints. Teplizumab took about 30 years to reach approval.
- No AI-designed molecule is in type 1 trials. That same review reported none of the repurposing candidates it surfaced were in a registered type 1 diabetes trial at the time of writing.
- “AI-designed” describes how something was made, not what it has been shown to do.
Share this
Listen to the Juicebox Podcast
Conversations about diabetes, five days a week, since 2015.
Read the full disclaimer
Two Different Roads Into Type 1
Two Different Roads Into Type 1
A genome study split type 1 diabetes by a genetic marker almost nobody has had tested. The two halves turned out to look less alike than expected.
Researchers at the University of Exeter and UC San Diego did something with type 1 genetics that had not been done before. Writing in Diabetologia on August 31, 2026, they took 9,091 people with type 1 and 14,157 people without it, sorted everyone by which high-risk HLA haplotype they carried, and ran the genetics separately in each group. The paper is open access.
Two haplotypes do most of the heavy lifting in type 1 risk. They are called DR3 and DR4, and by the researchers' own accounting roughly 90% of people of European ancestry who develop type 1 carry one or both. Which one a person carries tracks loosely with which islet antibody shows up first, and with how old they are at diagnosis.
Compared side by side, the two groups shared less genetic ground than the researchers expected. That result did not appear out of nowhere. It has been assembling for about a decade, in three other places first.
How big a gap is 0.68?
A genetic correlation of 1.0 would mean the measured genetic effects in the two groups line up almost perfectly. The team ran two comparisons as a yardstick. Splitting the same people by sex gave 0.88. Splitting them by whether they were diagnosed before or after age 8 gave 0.96 — close to no genetic difference at all.
Against those, 0.68 is a real separation, and the paper notes it sits near the correlation between Crohn's disease and ulcerative colitis. Worth being precise: the gap against the age split cleared statistical significance, and the gap against the sex split fell just short of it. Worth naming plainly: this work had industry funding alongside its university, government and nonprofit support. It received research funding from Randox Laboratories, and the lead author’s PhD studentship is funded by Randox — a diagnostics company that two of the authors hold a separate grant with, to develop a type 1 genetic risk score biochip. Several authors also disclose consulting or honoraria relationships with pharmaceutical companies.
One locus separated cleanly. Variants near IL2, a gene central to T cell activation, had a larger effect in DR4 carriers than in DR3 carriers, and it was the only locus to survive correction for multiple testing. Three others showed weaker signals that did not. Where the pattern gets broader is in which cellular machinery each group's variants landed in — and putting three groups side by side, a gradient appears.
The question underneath
There is a tension inside this paper worth sitting with. The framework most of this work rests on defines its two groups by age at diagnosis. But when this study split people that way, the genetic correlation came back at 0.96 — the two age groups looked nearly identical. Splitting by HLA produced the wider gap.
Which raises a question the field has not settled: which measurement is doing the real work?
This is the framework built from pancreas tissue. Children diagnosed before 7 show one pathology; those diagnosed after 12 show another. That evidence is direct rather than inferred — someone looked at the tissue. Age is also free, universal, and already sitting in every chart, which makes it usable in a way a genetic test is not.
An editorial read of the new findings would flip the order. Age at diagnosis may describe how fast the process ran rather than which process ran. HLA type is fixed at conception, never drifts, cannot go transiently undetectable the way an antibody can, and in this study separated the genetics more sharply than age did. The paper's own recommendation is that future trials consider DR3 and DR4 status in their design.
The researchers are direct about the limits, and their list is the most useful part of the paper. HLA type is only a partial stand-in for which antibody appears first. Everyone carrying both DR3 and DR4 was left out entirely, because they cannot be assigned to one group — and that combination carries the highest risk of all. The study ran only in people of European ancestry, so how well the findings generalize to other ancestry groups is not yet known. And the mast cell result rests on a single comparison that barely cleared its threshold.
Almost nobody living with type 1 knows their HLA type. It does not appear on a discharge summary, it never changes, and it may turn out to matter.
An editorial observationWhat this changes right now
Nothing, for anyone's day-to-day management. HLA typing is not part of routine type 1 care, no treatment decision currently depends on it, and nothing in this paper is approved as a way to choose a therapy. Insulin, pump settings, and CGM alarms are untouched by any of this. It is a study about where the disease comes from, not about how to manage it.
Where it could land is in trial design. If prevention studies start recording HLA background and reporting results by group, the field would learn faster why a drug helps one family and not another — which is the authors' own recommendation.
There is also a smaller group with a more immediate stake: the roughly 10% who carry neither haplotype, are diagnosed later, have fewer detectable antibodies, and sometimes get labeled type 2 for years before anyone revisits it. Anyone who has wondered whether their own diagnosis got sorted into the wrong bin has a reasonable question to raise — not a demand for a genetic test, but a conversation about what was measured and when. That is a conversation to have with a doctor who knows the whole history, not a decision to make from a study.
The main study. Luckett AM, McGrail C, Murrall K, et al. Genetic association stratified by HLA-DR3 and HLA-DR4 status reveals heterogeneity in pathways of progression to type 1 diabetes. Diabetologia, 31 August 2026. Open access. Funded by university, government and nonprofit sources, plus research funding from Randox Laboratories; the lead author holds a Randox PhD studentship, and two authors hold a Randox grant to develop a type 1 genetic risk score biochip. Authors separately disclose consulting and honoraria relationships with several pharmaceutical companies.
People with neither haplotype. McGrail C, Chiou J, Elgamal R, et al. Genetic discovery and risk prediction for type 1 diabetes in individuals without high-risk HLA-DR3/DR4 haplotypes. Diabetes Care 2025;48(2):202–211.
Antibody order in children. Krischer JP, Lynch KF, Schatz DA, et al. The 6 year incidence of diabetes-associated autoantibodies in genetically at-risk children: the TEDDY study. Diabetologia 2015;58(5):980–987.
Pancreas tissue and age at diagnosis. Leete P, Oram RA, McDonald TJ, et al. Studies of insulin and proinsulin in pancreas and serum support the existence of aetiopathological endotypes of type 1 diabetes associated with age at diagnosis. Diabetologia, 2020.
Mast cells in donated pancreas tissue. Martino L, Masini M, Bugliani M, et al. Mast cells infiltrate pancreatic islets in human type 1 diabetes. Diabetologia 2015;58(11):2554–2562.
IL2 and age at diagnosis. Howson JMM, Cooper JD, Smyth DJ, et al. Evidence of gene-gene interaction and age-at-diagnosis effects in type 1 diabetes. Diabetes 2012;61(11):3012–3017.
Teplizumab subgroups. Herold KC, Bundy BN, Long SA, et al. An anti-CD3 antibody, teplizumab, in relatives at risk for type 1 diabetes. New England Journal of Medicine 2019;381(7):603–613. Subgroup findings were prespecified but not adjusted for multiple comparisons.
How to read this evidence. Every study above is a genetic association or tissue study, not a test of a treatment. They describe patterns across large groups of people; none of them predicts what will happen to any one person. Association is not cause: finding that a variant is more common in a group does not show that it drives the disease. The teplizumab subgroup numbers come from prespecified analyses that were not adjusted for multiple comparisons and have been described in the literature as exploratory. HLA typing is not part of standard type 1 care, and no therapy is approved to be selected on the basis of DR3 or DR4 status.
This post is for educational purposes only and is not medical advice. Nothing here is a recommendation to start, stop, or change any medication. Talk with your doctor before making any changes to your care.
Two Different Roads Into Type 1
A genome study split type 1 by HLA background and found the two groups less genetically alike than expected. It is a study about where the disease comes from, not about how it is managed, and nothing in it changes anyone’s care.
What the research shows
- Luckett et al., Diabetologia, August 2026: 9,091 people with type 1 and 14,157 without, sorted by whether they carried the DR3 or the DR4 haplotype. Setting the HLA region aside, the two groups’ genetic effects correlated at 0.68.
- Splitting the same people by age at diagnosis gave 0.96, and by sex 0.88. The gap against the age split cleared statistical significance; the sex gap fell just short.
- DR4 risk variants clustered in T cell machinery. DR3 variants clustered in mast cell regulatory regions and pathways for secretion and cellular stress.
- IL2 was the only locus whose difference between groups survived correction for multiple testing.
- A companion study (McGrail et al., Diabetes Care, 2025) found the roughly 10% who carry neither haplotype lean toward antigen presentation, innate immunity and beta cell pathways instead.
What it does not show
- Not a new diagnosis, subtype or test. No category changed and no one’s care changes because of it.
- Everyone carrying both DR3 and DR4 was excluded, because they cannot be assigned to one group — and that combination carries the highest risk of all.
- The study ran only in people of European ancestry, so how well the findings generalize to other ancestry groups is not yet known.
- The mast cell result rests on a single comparison at p=0.043. A lead, not a finding.
- These are association studies. Finding a variant more often in a group does not show that it drives the disease.
If you live with type 1
- HLA typing is not part of routine type 1 care, and no treatment decision currently depends on it.
- The likely near-term consequence is trial design — recording HLA background and reporting results by group, which is the authors’ own recommendation.
- If your own diagnosis was ever unclear — later onset, few detectable antibodies, an initial type 2 label — that is a reasonable thing to raise with the doctor who knows your whole history.
Funding. The main study had industry funding alongside university, government and nonprofit support, including research funding and a PhD studentship from Randox Laboratories.
Educational purposes only and not medical advice. Talk with your doctor before making any changes to your care. — juiceboxpodcast.com
Share this
Listen to the Juicebox Podcast
Conversations about diabetes, five days a week, since 2015.
Read the full disclaimer
Old Mice Lived Longer on Semaglutide
Old Mice Lived Longer on Semaglutide. Look at What the Drug Copied.
A Nature study started 20-month-old female mice on daily semaglutide and they outlived the untreated mice by about 12 percent. The more interesting finding is that the drug’s gene signature overlapped with calorie restriction — a similar calorie-restriction signal to the one that showed up inside beta cells in a separate GLP-1 study.
Start an old mouse on semaglutide, keep it on the drug for the rest of its life, and it lives about 12 percent longer. That is the headline from a study published in Nature on September 2 by Danica Chen’s lab at the University of California, Berkeley. Median lifespan went from 742 days in the untreated mice to 834 days in the treated ones. In mouse time, that is about three months.
The mice were 20 months old when the injections started. One outside expert who reviewed the paper put that at roughly the equivalent of a 62-year-old woman. They were all female. And they ate 24 percent less than the untreated mice, which is the detail the rest of the story turns on.
Because eating 24 percent less is, on its own, a well-known way to make a mouse live longer. Researchers call it calorie restriction, and it is already known to extend lifespan in many animal models. So the real question in this paper is not whether the mice lived longer. It is whether semaglutide did anything beyond making them eat less. The authors say it did. Here is the study, one step at a time.
What the study shows
This work was funded by the National Institute on Aging, part of the NIH, and by the National Institute of Food and Agriculture. No drug company is listed as a funder, and Novo Nordisk, which makes semaglutide, was not a sponsor. The paper does disclose that the Regents of the University of California filed a patent application covering GLP-1 receptor agonists for healthy ageing.
On the evidence ladder this is animal research: well run, blinded where it could be, with 79 mice in the lifespan arm and 10 per group in the function arms. It shows what the drug did in old female laboratory mice of a single inbred strain. It cannot show what the drug does to a person’s lifespan, and the authors say so. Whether GLP-1 activation changes aging in humans, they write, will take long-term clinical studies designed to measure aging outcomes in older people.
The comparison to calorie restriction is where the paper earns its place, so it gets its own box.
Now the part worth holding up against something else. A separate post on this blog covered a Salk Institute study, published in PNAS in March, that traced how a GLP-1 changes gene expression inside the beta cell. Buried in that paper was a line that was easy to pass over at the time: in mouse islets, the drug pushed beta cells toward the marker genes seen in calorie-restricted mice, and away from the ones seen in mice on a high-fat diet.
Different lab. Different organ. Different question. A similar signal. The Salk team was looking at one cell type in the pancreas, using prolonged exposure to a different GLP-1 in isolated mouse islets. The Berkeley team was looking at the liver, blood, brain, and lifespan of a whole animal, and at a broader aging program. These are not the same experiment and not the same gene list. Neither paper cites the other, and nothing in either one proves they are describing the same mechanism. That connection is an editorial observation, not a finding. But when two groups studying GLP-1 signaling from opposite ends both end up holding something that looks like a calorie-restriction signature, that is worth writing down.
What nobody knows yet
The paper is clear about its edges. Two of them decide how much of this transfers to anyone reading it.
OPEN QUESTION 1
Does it hold in males, or in people?
Every mouse in this study was female, chosen to avoid the fighting and injuries that complicate long-term studies in male mice. Aging differs by sex, in mice and in people. All of the mice were one inbred strain, bred for genetic uniformity, which is the opposite of a human population. On the human side, one outside expert noted that the mice started the drug at about 75 percent of their expected lifespan and stayed on it for life; for a woman in the UK that would mean starting around 62 and never stopping. Large human trials have linked GLP-1 drugs to fewer heart attacks, strokes, and deaths in people with type 2 diabetes and with obesity, but that is what those trials counted. They did not measure aging. Nobody has run the human version of this experiment.
OPEN QUESTION 2
Is it the drug, or is it the eating less?
The authors argue for both. Most of the benefit matched what a plain 24 percent food cut produced, which points at the calories. But three outcomes improved beyond what calorie restriction achieved, and the treated mice never went through the fasting-and-foraging rhythm that calorie restriction imposes. The authors read that as a sign that GLP-1 receptor activation taps something calorie restriction does not. An outside neuroscientist offered a simpler explanation for the brain results: the semaglutide mice moved more, and physical activity on its own is known to boost the growth of new neurons. The paper cannot settle which is right, and it does not claim to.
Twelve percent longer in a mouse is a result. Twelve percent longer in a person is a story until someone runs the trial.
Where this standsWhat this means if you live with diabetes
The arithmetic is sitting right there and everyone is going to do it. Twelve percent of an 80-year life is nine or ten years. Nobody has earned that number. Mouse lifespan gains shrink, sometimes to nothing, on the way to people, and the mice in this study had no diabetes, no autoimmunity, and no other medications on board.
If you have type 2 diabetes and take a GLP-1, this study does not change your reasons for taking it. It adds a mechanism story to benefits that human trials had already reported for hearts and kidneys. If you have type 1, the picture is different. No GLP-1 is approved for type 1 diabetes as of this writing; some people use one off-label with a doctor’s help, and this paper says nothing about safety, insulin needs, or DKA risk in that setting. It is a study of aging, not a study of diabetes.
The takeaway, and this is an editorial read, not the paper’s: the human evidence for these drugs that matters most was never about the scale. It was about the things that end lives early: heart attacks, kidney failure, strokes. This study offers one plausible reason those benefits show up: semaglutide pushed tissues toward some of the same biological states produced by calorie restriction. Because the treated mice also ate 24 percent less, the two effects cannot be cleanly separated — although several outcomes suggest GLP-1 receptor activation may do something beyond reduced calorie intake alone. Whether that adds years is unknown. Whether it removes some of the wear along the way is the more useful question, and it is the one the human trials are built to answer. If any of this touches your care, bring it to a doctor who knows your diabetes and knows you.
The study. Feng Y, Barthez M, Wang Y, Chen Y, Qiu H, Wang CL, Heydari K, Delcroix M, Rasmussen LJ, Bohr VA, Chen D. Late-life semaglutide treatment slows ageing and extends lifespan in female mice. Nature, published September 2, 2026 (PubMed 42686906). Funding listed: National Institute on Aging (R01AG063404, R01AG063389, R01AG082105) and the National Institute of Food and Agriculture. No industry funding listed. Competing interests declared in the paper: the Regents of the University of California filed patent application 64/113,481, covering GLP-1 receptor agonists for healthy ageing. Evidence tier: animal research — 20-month-old female C57BL/6 mice, 39 saline and 40 semaglutide in the lifespan arm, 10 per group in the five-month comparison.
Institutional release. UC Berkeley, Department of Metabolic Biology & Nutrition, GLP-1 treatment extends the lifespan of older, healthy mice, September 2, 2026. Used for the funding statement and author affiliations.
Outside expert comment. Science Media Centre, expert reaction, September 2, 2026. The age-62 comparison is from Dr. Laura Sinclair, University of Exeter. The exercise explanation for the brain findings is from Prof. Tara Spires-Jones, University of Edinburgh. Prof. Naveed Sattar, University of Glasgow, placed the study alongside human trial evidence on mortality; he declares consulting and speaking fees from Novo Nordisk and other manufacturers. The human outcome trials referred to above are the cardiovascular outcome studies of these drugs, for example LEADER (liraglutide in type 2 diabetes, NEJM 2016) and SELECT (semaglutide in overweight and obesity without diabetes, NEJM 2023), which measured heart attacks, strokes, and deaths, not aging.
The related beta-cell study. Van de Velde S, Yu J, Evensen KG, Pakhlevanyan E, Williams AE, Shaw RJ, Montminy M. Med14 phosphorylation shapes genomic response to GLP-1 agonists. Proceedings of the National Academy of Sciences 123(10): e2536772123, March 4, 2026. That study used prolonged exendin-4 exposure in isolated mouse islets and found induction of genes enriched in calorie-restricted beta-cell states; the Nature study used semaglutide in whole animals and compared a broader aging and calorie-restriction program. They are related signals, not the same experimental result. The Juicebox post on the beta-cell study is here. The connection between the two papers is an editorial observation; neither paper cites the other.
Type 1 approval status. No GLP-1 is approved for type 1 diabetes. Two phase 3 trials involving semaglutide in type 1 are under way, and they are testing different things. NCT06894784 (SEMPA, McGill University Health Centre, 36 adults) is a randomized, double-blind, 2×2 factorial crossover trial of semaglutide and empagliflozin, alone and combined, added to automated insulin delivery, with time in range as the primary endpoint. NCT06082063 (Steno 1, Steno Diabetes Center Copenhagen, 2,000 participants) is a multifactorial cardiovascular-risk trial in high-risk adults with type 1, in which semaglutide is one of several agents assigned by individual risk profile alongside sotagliflozin, finerenone, ezetimibe and PCSK9 inhibitors; it is not a test of semaglutide on its own. Both checked September 2026.
How to read this evidence. This is a lifespan and physiology study in one inbred strain of laboratory mouse, female only, treated with one drug at one dose starting late in life. It shows what happened to those animals. It cannot show whether semaglutide lengthens, shortens, or does nothing to a human life, and it did not study anyone with diabetes. Semaglutide (Ozempic, Wegovy) is FDA-approved for type 2 diabetes and weight management; it is not approved for type 1 diabetes.
This post is for educational purposes only and is not medical advice. Nothing here is a recommendation to start, stop, or change any medication. Talk with your doctor before making any changes to your care.
Listen to the Juicebox Podcast
Conversations about diabetes, five days a week, since 2015. Search “GLP-1” in the Juicebox FAQ for the episodes on off-label use in type 1.
Read the full disclaimer