Smart Insulin and Oral Delivery: Beyond the Needle
Insulin Biosimilars
3.2x
The fold-increase in insulin receptor affinity that Novo Nordisk's glucose-sensitive insulin NNC2215 achieves when blood glucose rises from 3 to 20 mM, demonstrated in preclinical models.
Hoeg-Jensen et al., Nature, 2024
Hoeg-Jensen et al., Nature, 2024
View as imageInsulin has been injected since 1922. That discovery saved millions of lives, but a century later, the fundamental delivery method has barely changed: a needle, a syringe or pen, and a patient who must calculate how much to take and when. Smart insulin, a class of glucose-responsive analogs and delivery systems designed to release insulin only when blood sugar rises, represents the most ambitious attempt to change that equation. For a broader look at how insulin formulations have evolved, see our complete guide to insulin biosimilars.
The parallel pursuit of oral insulin delivery has consumed decades of pharmaceutical research with no commercial product to show for it. Peptides are notoriously fragile in the gastrointestinal tract, and insulin is no exception. But recent advances in nanoparticle engineering, gastric auto-injectors, and cell-penetrating peptide platforms have pushed oral bioavailability from the low single digits toward numbers that might actually work in patients.
This article maps both frontiers: where glucose-responsive insulin stands after its first published preclinical proof-of-concept, and how close oral delivery technologies are to replacing the needle.
Key Takeaways
- NNC2215, a glucose-sensitive insulin conjugate from Novo Nordisk, showed a 3.2-fold increase in insulin receptor affinity when glucose rose from 3 to 20 mM in preclinical studies (Hoeg-Jensen et al., Nature, 2024)
- Gastric auto-injector capsules delivered recombinant human insulin with up to 80% bioavailability in swine, reaching peak plasma concentration within 30 minutes[1]
- Oral insulin-peptide nanocomplexes using cell-penetrating peptides enhanced intestinal translocation more than 22-fold and controlled blood glucose in diabetic mice over multiple days[2]
- Machine learning models can now predict peptide stability in simulated gastric fluid with 75.1% accuracy, potentially accelerating oral insulin formulation design[3]
- Inhaled insulin (Exubera) was approved in 2006 but pulled from the market within a year due to poor sales, not safety failures[4]
- GLP-1 receptor agonists and dual GIP/GLP-1 agonists like tirzepatide are already replacing some insulin regimens, reducing HbA1c by an additional 1.08% versus long-acting insulin[5]
What Is Smart Insulin?
Smart insulin refers to any insulin formulation or analog whose activity adjusts automatically based on the patient's blood glucose concentration. The concept mimics what healthy pancreatic beta cells do naturally: release insulin in proportion to glucose levels, then stop when blood sugar drops.
Three broad strategies have been explored. The first uses glucose oxidase (GOx), an enzyme that converts glucose to gluconic acid, creating a local pH change that triggers insulin release from a polymer matrix. The second relies on glucose-binding proteins like concanavalin A (ConA), which competitively release insulin when glucose concentrations rise. The third, and most recent, involves direct chemical modification of the insulin molecule itself so that its receptor-binding affinity changes in response to glucose.
All three aim for the same outcome: insulin that cannot cause hypoglycemia because it becomes inactive when blood sugar is already low.
NNC2215: The First Glucose-Sensitive Insulin to Reach Preclinical Proof
In October 2024, Novo Nordisk published the first demonstration of an insulin conjugate with intrinsic glucose sensitivity in the journal Nature. NNC2215 is a chemically modified insulin attached to a glucose-binding macrocycle. When blood glucose is low (3 mM), the macrocycle occupies the insulin's receptor-binding site, blocking its activity. When glucose rises (20 mM), glucose displaces the macrocycle, and the insulin molecule regains 3.2 times its baseline receptor affinity.
In diabetic rat models, NNC2215 attenuated hypoglycemia compared to conventional insulin while still covering glucose excursions after meals. The insulin also remained stable for weeks without refrigeration before first use.
NNC2215 has not entered human trials. Novo Nordisk has stated that further optimization of its pharmacological properties is ongoing. The gap between a preclinical demonstration and a clinically viable product is wide. Earlier glucose-responsive approaches using GOx and ConA have been explored since the 1980s without producing an approved product, largely because of immunogenicity, slow response kinetics, and manufacturing complexity.
Why Oral Insulin Delivery Is So Difficult
Insulin is a 51-amino-acid peptide hormone. When swallowed, it faces three barriers that injectable delivery avoids entirely.
Enzymatic degradation. Pepsin in the stomach and trypsin and chymotrypsin in the small intestine break peptide bonds within minutes. Insulin's half-life in simulated gastric fluid is measured in seconds to minutes depending on pH and enzyme concentration. Machine learning models trained on 109 peptide incubation experiments found that lipophilicity, rigidity, and molecular size are the key determinants of whether a peptide survives the GI tract.[3]
Mucosal permeability. Even if insulin survives digestion, the intestinal epithelium is a tight barrier to molecules above 500 Da. Insulin, at roughly 5,800 Da, does not cross passively. Permeation enhancers or active transport mechanisms are required. Multi-unit particulate formulations using layered permeation enhancers and pH-responsive coatings have shown improved bioavailability in dog models when disintegration occurs at higher intestinal pH.[6]
First-pass metabolism. Oral drugs pass through the liver before reaching systemic circulation. The liver degrades a significant fraction of absorbed insulin, further reducing what reaches target tissues. This is actually the route endogenous insulin normally takes (the pancreas delivers insulin directly into the portal vein), so some researchers argue oral delivery could produce a more physiological insulin profile than subcutaneous injection.
Oral Insulin Candidates That Reached Clinical Trials
Two oral insulin programs advanced furthest before encountering problems.
ORMD-0801 (Oramed Pharmaceuticals) used a protease inhibitor and absorption enhancer combination to protect insulin through the GI tract. In a 12-week dose-finding Phase 2 trial, the 8 mg twice-daily dose showed some reduction in HbA1c versus placebo in people with type 2 diabetes, but the effect was modest and variable. Bioavailability was approximately 7%. The Phase 3 program was terminated in January 2023 after failing to meet primary endpoints. The drug was safe and well-tolerated, with no hypoglycemia or weight gain, but it did not lower blood sugar enough to be clinically meaningful.
Tregopil (Biocon) took a different approach: an ultra-fast oral prandial insulin using an absorption enhancer that achieved peak blood concentrations within 15 to 20 minutes of dosing. In a Phase 2/3 study comparing Tregopil to injectable insulin aspart in type 2 diabetes, Tregopil matched aspart for early postprandial glucose control but was inferior for late postprandial effects. HbA1c did not improve at 24 weeks compared to the injected comparator. Bioavailability ranged from 18 to 28%, substantially better than ORMD-0801 but still insufficient for reliable glucose management.
Neither program produced a commercial product. The pattern suggests that chemical permeation enhancement alone is unlikely to solve oral insulin delivery.
Nanoparticle and Peptide-Based Oral Delivery Systems
The most promising recent advances use engineered delivery vehicles rather than simple absorption enhancers.
Gastric auto-injectors. Abramson and colleagues at MIT, working with Novo Nordisk, developed an orally dosed liquid capsule that physically injects its payload into the stomach or intestinal wall. In swine studies, the device delivered recombinant human insulin with up to 80% bioavailability, reaching maximum plasma concentration within 30 minutes. This represents a one to two order-of-magnitude improvement over chemical permeation enhancement approaches.[1] The same capsule also delivered a GLP-1 analog, adalimumab, and epinephrine across multi-day dosing experiments.
Cell-penetrating peptide (CPP) nanocomplexes. Rehmani et al. (2023) conjugated insulin with a modified CPP platform called GET (Glycosaminoglycan-binding-Enhanced-Transduction) to form nanocomplexes approximately 140 nm in size. These complexes enhanced insulin translocation across differentiated Caco-2 intestinal epithelium models by more than 22-fold compared to free insulin. Epithelial cells accumulated the nanocomplexes and acted as depot reservoirs for sustained insulin release. In diabetic mice, serial oral dosing controlled elevated blood glucose over several days.[2]
Both approaches remain in preclinical development. The gastric auto-injector faces manufacturing and cost challenges for a drug taken daily. The CPP nanocomplex approach needs to demonstrate scalability and long-term safety. For more on how cell-penetrating peptides are being used across drug delivery, see articles on CPP cargo conjugates.
Inhaled Insulin: What Exubera Taught the Field
Inhaled insulin was supposed to be the first needle-free insulin revolution. The lungs offer a large surface area, rich vascularity, and thin epithelial barriers. Exubera (Pfizer/Nektar) won FDA approval in January 2006 as the first inhaled insulin for adults with type 1 or type 2 diabetes.[4]
Clinical trials showed Exubera was effective. It matched subcutaneous mealtime insulin for glycemic control, and patients preferred it. But the device was large and cumbersome (roughly the size of a tennis ball can), the pricing was high, and pulmonary function monitoring was required. Pfizer pulled it from the market in October 2007, absorbing a $2.8 billion write-off. Sales in its final quarter were just $12 million.
Afrezza (MannKind), a smaller-device inhaled insulin approved in 2014, remains on the market but captures a tiny fraction of the insulin market. The pulmonary route does work for peptide delivery.[7] The failure was commercial, not scientific. Patients and prescribers were not willing to trade injection convenience for a bulky inhaler with uncertain long-term lung effects.
How GLP-1 and Dual Agonists Are Changing the Landscape
While researchers chase needle-free insulin, a different class of peptides is making some insulin regimens unnecessary altogether.
GLP-1 receptor agonists and the newer dual GIP/GLP-1 agonists stimulate insulin secretion only when blood glucose is elevated, providing a form of "smart" glucose-dependent control without modifying the insulin molecule itself. Multiple peptide therapies can now replace or supplement insulin injection regimens in type 2 diabetes. For a detailed comparison, see our guide to rapid-acting vs long-acting insulin analogs.
Tirzepatide, a dual GIP/GLP-1 receptor agonist, demonstrated striking superiority over long-acting insulin in a meta-analysis of three Phase 3 trials encompassing 4,339 patients. Compared to insulin glargine and degludec, once-weekly tirzepatide reduced HbA1c by an additional 1.08%, body weight by 10.61 kg, and cut the risk of hypoglycemia by 54%.[5]
A separate meta-analysis of eleven trials found that adding GLP-1 receptor agonists or tirzepatide to basal insulin improved HbA1c by 1.0% and reduced body weight by 3.95 kg without increasing hypoglycemia risk.[8] The combination of basal insulin with a GLP-1 agonist, such as the fixed-ratio iGlarLixi formulation, also improved beta-cell glucose sensitivity by 35% over 26 weeks.[9]
These results mean the target population for novel insulin delivery technologies may be shrinking. Many type 2 diabetes patients who once required multiple daily insulin injections can now achieve better control with a weekly peptide injection. Non-insulin pharmacological options for type 1 diabetes, while still limited, are also expanding.[10]
For context on how these dual agonists work, see how tirzepatide's dual mechanism differs from single GLP-1 agonists and the SURMOUNT trials evidence.
Where the Science Stands in 2026
The honest assessment: smart insulin is years away from patients. NNC2215 is the only glucose-responsive insulin analog with published preclinical data showing the concept works in living animals. It has not entered human trials. Every previous glucose-responsive delivery system (GOx microneedles, ConA hydrogels, phenylboronic acid polymers) has stalled in preclinical or early clinical stages.
Oral insulin is closer to reality but still without a clear winner. The chemical permeation enhancement approach (ORMD-0801, Tregopil) has effectively failed in late-stage trials. The mechanical approaches (gastric auto-injectors) and biological approaches (CPP nanocomplexes) show dramatically better bioavailability numbers but are years from human efficacy data.
The technologies most likely to change insulin delivery in the near term are not novel insulin molecules at all. They are the incretin-based peptides that reduce or eliminate the need for insulin in many patients with type 2 diabetes, and the closed-loop automated insulin delivery systems (artificial pancreas technology) that use existing insulin formulations with algorithmic dosing.
The long game for smart insulin and oral delivery remains scientifically sound. A glucose-responsive insulin that cannot cause hypoglycemia would be transformative for type 1 diabetes, where incretin agonists cannot replace insulin. An oral insulin with reliable bioavailability would remove the single biggest barrier to early insulin initiation in type 2 diabetes. Both goals are worth pursuing. Neither is imminent.
The Bottom Line
Smart insulin technology achieved its first preclinical proof-of-concept in 2024 with NNC2215, a glucose-sensitive insulin conjugate that adjusts receptor affinity based on blood sugar levels. Oral insulin delivery has advanced through gastric auto-injectors and peptide-based nanocomplexes showing 80% and 22-fold improved bioavailability respectively, but no oral insulin product has reached market. Meanwhile, GLP-1 and dual GIP/GLP-1 agonists are already replacing insulin regimens for many type 2 diabetes patients, potentially narrowing the population that needs novel insulin delivery the most.