Concept: sub-20-minute glucose-responsive mealtime insulin

This is not a product — it is a target, and one of the clearest statements of what the Type 1 field should be building. It is a mealtime (prandial / bolus) insulin concept: the dose you would take to cover a meal, the role today filled by rapid and ultra-rapid analogs.³ Two properties define it. First, speed: begin working within a few minutes and reach peak effect in under 20 minutes. Second, ideally, glucose-responsiveness ("smart" insulin): a molecule that stays inert until glucose rises, then activates — a built-in safety net against lows.³

Why these numbers? A healthy pancreas releases a rapid first-phase insulin burst within roughly 1-2 minutes of glucose rising.⁷ By contrast, today's fastest approved analog (faster aspart, Fiasp) only reaches its glucose-lowering peak at about 90-133 minutes and tails for 5-7 hours.⁶ That multi-hour gap between eating and insulin action is the heaviest physical lag an automated insulin-delivery system has to predict around — and the reason hybrid closed-loop systems still ask you to announce meals. A sub-20-minute insulin would shrink that lag three- to five-fold.⁶

PK/PD target (aspirational). Onset: minutes. Time-to-peak: under 20 minutes. Duration / tail: short — fast clearance, or a glucose-responsive shut-off that ends action as glucose normalizes. There is no published human PK/PD curve, because no such insulin exists yet; these are design goals, not measured values.³

Has anyone built it? Pieces, not the whole. MK-2640 (Merck) was the first glucose-responsive insulin tested in people: in 16 adults with type 1 diabetes, the intravenous analog failed to show a glucose-dependent effect, and it was about 25-fold less potent than regular insulin — the program did not advance.² In 2024, NNC2215 (Novo Nordisk, with Carbometrics) showed a genuine molecular glucose switch — insulin-receptor affinity rose 3.2-fold as glucose went from 3 to 20 mM, and it protected pigs and rats against hypoglycemia — but it has been tested only in animals.¹ Separately, glucose-responsive polymer and oral-reservoir systems have normalized glucose for a day in mice and pigs without causing lows, again preclinically.⁵ Authoritative reviews group all of this into three strategies — CGM-linked closed loops (already real), glucose-responsive polymers, and mechanism-based molecular modifications — and conclude that no intrinsic glucose-responsive insulin analog has reached the clinic or been approved.⁴

Absorption variability and exercise. A faster route would not automatically be a more consistent route — predictable day-to-day absorption is unproven for any sub-20-minute chemistry, and MK-2640's failure shows how hard the consistency bar is.² The flip side is the prize: a truly glucose-responsive insulin would, in principle, dial itself down as glucose falls during activity, which is exactly the hypoglycemia protection NNC2215 demonstrated in animals.¹

Delivery, approvals, access. Undefined — no regulator has reviewed such a product, and there is no price, region, or biosimilar to report.

What's coming. This is an active, funded frontier. The Type 1 Diabetes Grand Challenge (Diabetes UK, the Steve Morgan Foundation, and Breakthrough T1D) has committed £15 million specifically to ultra-rapid and glucose-responsive "smart" insulins, explicitly so closed-loop systems can react without manual meal and exercise entry.⁹ Breakthrough T1D (formerly JDRF) is funding glucose-responsive insulin work.¹⁰ This is an unranked design concept. Named insulins and development programmes are assessed in their own records.

Human studies versus the design goal. LY3938577 and NNC0363-1063 have entered human phase 1 studies, as covered in glucose-responsive insulin. Their existence does not establish the combined speed and glucose-responsive profile envisioned here; NNC2215’s reported switch remains animal-stage evidence.