Inreda AP (dual-hormone)

The Inreda AP is a dual-hormone artificial pancreas in real-world use. It delivers two hormones — insulin to lower glucose and glucagon to raise it — and runs as a genuinely fully closed loop: no carb counting, no meal announcements, and no exercise input.² That full automation is possible precisely because the second hormone gives the algorithm an active brake. Insulin-only systems can only stop delivering insulin and then wait; glucagon lets the Inreda AP actively push glucose back up, so its controller can run more aggressively without driving the user low.³

How the second hormone defends lows. When the device's sensors detect glucose falling, it delivers small doses of glucagon to mobilise stored glucose from the liver — a fast counter-regulatory response that approximates what a healthy pancreas does. In the 1-year real-world trial, median time below range was just 1.36% (well under the 4% safety target), and in an earlier randomized crossover study the system delivered glucagon automatically with no exercise or meal announcements.²³

Trial outcomes. The pivotal evidence is a 1-year, eight-centre, single-arm trial in 79 adults in the Netherlands, published in The Lancet Digital Health (2024). Time in range was 55.5% before the trial and 80.3% at 12 months — a difference of roughly six extra hours per day in target.² Diabetes distress (PAID) changed from 30 to 10 and well-being (WHO-5) from 60 to 76 over the trial.² An earlier randomized crossover trial in Diabetes Care found ~87% time in range on the closed loop versus ~54% on usual pump therapy.³ A separate JAMA Surgery trial in people with no pancreas at all (after total pancreatectomy) showed the same pattern — higher time in range with low reported hypoglycemia.⁴

The glucagon-stability challenge. Reconstituted glucagon (the device uses GlucaGen) is chemically unstable in solution and tends to aggregate, so there is no shelf-stable pre-filled cartridge yet — the user fills the glucagon reservoir by hand and replaces it every single day.⁵⁶ Solving this is a central formulation problem for two-hormone systems; newer stable analogs (e.g. dasiglucagon) were developed to address this kind of instability.⁶

Device burden. Two hormones mean roughly double the hardware: two infusion sets, two reservoirs, and two redundant CGM sensors (one drives therapy, one cross-checks it), all in a body-worn unit about 12 x 8 x 4 cm.⁵ Users in a qualitative study described real benefits (better glucose, "feeling more like a person without diabetes") but flagged the daily glucagon swap, multiple components, and alarms as the main hassles — reported as key reasons for discontinuation.⁷

Maturity and access. The current generation, AP5, is CE-marked under the EU Medical Device Regulation, the bar for genuine clinical use.⁸ But access is narrow: it is available essentially only in the Netherlands, through Inreda-run projects and limited insurer coverage, with no FDA clearance and no broad commercial EU sale.⁵⁸ We mark it available as a regional (EU) deployment, not a globally purchasable product.

What's coming. Inreda is developing the next-generation AP6, aimed at a smaller, lighter device.⁸ The field-wide priorities are a stable, pre-fillable glucagon to end the daily-refill burden, pediatric use (a youth/adolescent study is described as underway), and broader reimbursement to move beyond a single country. The Inreda AP is an existence proof that fully closed, no-announcement automation is real — the remaining work is shrinking the box and stabilising the second hormone.²⁶

PANORAMA, a different diabetes population. A published ADA 2026 abstract reports 34 adults randomized after total pancreatectomy, with 20 completing both periods for the predefined primary analysis. Three-month TIR was 81% with the bihormonal system versus 57% with usual care (difference 24 percentage points; 95% CI 16–32). Time below 70 mg/dL was 0.9% versus 1.2%; its difference was not statistically established. This supports further evaluation after pancreatic surgery, not an interchangeable result in autoimmune T1D or a comparison against modern hybrid AID.¹