Oral peptide delivery — the engineering breakthroughs that may transform peptide therapy

The reason is chemistry. Peptides are strings of amino acids held together by peptide bonds, and the gastrointestinal tract is exceptionally good at cleaving those bonds. The stomach's acidic environment alone degrades many peptide sequences before they reach the small intestine. Proteases in the intestinal lumen continue the work. The intestinal epithelium — even for peptides that survive to reach it — presents a permeability barrier designed to keep large hydrophilic molecules out of the bloodstream. For most peptides, the result of swallowing rather than injecting is that essentially nothing makes it to circulation. Bioavailability of one or two percent is typical. For a drug that's biologically active at nanomolar concentrations, that's not a rounding error — it's the difference between a functional dose and nothing.

This is why what Novo Nordisk accomplished with oral semaglutide — approved by the FDA in 2019 under the brand name Rybelsus — represents something more significant than a line extension. Semaglutide injected produces predictable, controllable plasma concentrations at doses in the milligram range. Semaglutide swallowed, without engineering, produces essentially no clinically meaningful concentration. Getting from one to the other required solving a specific delivery problem, and the solution — SNAC technology — is worth understanding in some detail.

SNAC stands for sodium N-[8-(2-hydroxybenzoyl)amino] caprylate. It's a permeation enhancer: a molecule coformulated with semaglutide that accomplishes several things simultaneously. SNAC locally raises the pH around the tablet as it dissolves in the stomach, reducing the activity of pepsin and other gastric proteases. It also transiently increases the permeability of gastric epithelial cells — not the intestinal epithelium, which most permeation enhancer work has targeted, but the stomach itself, which is less enzymatically hostile. The result is a brief window during which semaglutide can cross the gastric wall and enter the bloodstream before the intestinal proteases have a chance to finish it off. The bioavailability that results is still low by small-molecule standards — roughly one percent — but because the drug is active at very small doses and the tablet is formulated with this in mind, that one percent is enough to produce a clinically meaningful plasma concentration. The pill works. It works differently than the injection, with more inter-individual variability in absorption and some sensitivity to food timing, but it works.

Rybelsus is now an approved diabetes medication, and the oral semaglutide program has expanded: Wegovy in injection form is approved for obesity, and an oral semaglutide at higher doses for obesity is progressing through clinical development. But SNAC is specific to semaglutide and to gastric absorption — it doesn't represent a universal solution. Different peptides have different chemical properties, different protease vulnerabilities, and different target pharmacokinetics, and each oral delivery challenge tends to require its own engineering.

The broader field has taken several different approaches. Lipid nanoparticles — the same delivery technology that carried mRNA vaccines into cells — are being explored as peptide carriers. The nanoparticle encapsulates the peptide, protecting it from proteases, and can be engineered for surface properties that influence where in the GI tract it releases its payload and how it interacts with epithelial cells. Early research suggests this approach can meaningfully improve the oral bioavailability of some peptides, though the work of getting from "meaningful improvement" to "clinically sufficient" varies enormously by compound. The mRNA vaccine experience accelerated investment in lipid nanoparticle manufacturing and formulation science, and some of that infrastructure is now being applied to peptide delivery — a secondary benefit of a platform technology that was developed for a completely different purpose.

Ionic liquid platforms represent another direction. Ionic liquids are salts that are liquid at room temperature, and certain formulations of peptides in ionic liquid matrices have shown improved permeation across biological membranes in preclinical models. The idea is to use the ionic liquid to solubilize the peptide in a form that interacts differently with the lipid bilayers of epithelial cells. Some of this work has produced striking results in animal models; whether those results translate to humans and at what cost is still being worked out.

Perhaps the most dramatic approach — and the one that most clearly illustrates how seriously the field is taking the oral delivery problem — is the SOMA capsule developed at MIT. SOMA stands for self-orienting millimeter-scale actuator, and it's essentially a small robotic device in pill form. The capsule is designed with a weighted geometry that causes it to orient itself against the stomach wall after swallowing, then deploy a small microneedle to inject the peptide directly through the gastric mucosa into the submucosa or bloodstream. The needle is made from compressed drug — it dissolves after injection. The device itself passes harmlessly through the GI tract. In porcine models, SOMA delivered insulin and other peptides with plasma concentrations comparable to subcutaneous injection. Human trials are a longer road, but the concept demonstrates that the engineering imagination being applied to this problem has genuinely expanded.

Buccal and sublingual delivery are more modest alternatives that don't require the same engineering sophistication. Some peptides — oxytocin, for example — are small enough and lipophilic enough that sublingual or buccal absorption can be meaningful, bypassing first-pass GI degradation by delivering directly to the bloodstream through the highly vascularized oral mucosa. These routes work for some compounds and not others; the molecular weight and hydrophilicity constraints are tight. But for the peptides where they work, they represent a meaningfully simpler delivery option than injection.

Beyond delivery engineering, peptide chemistry itself is being modified to improve oral viability. Cyclic peptides — peptide sequences that form ring structures rather than linear chains — are substantially more resistant to proteolytic degradation than their linear counterparts because proteases have a harder time accessing and cleaving the peptide bonds. Some cyclic peptide drugs are orally bioavailable at levels that linear analogs of similar molecular weight aren't. Cyclosporine is the classic example: it's a cyclic peptide with meaningful oral bioavailability, which is why it became a viable oral immunosuppressant. Newer drug discovery programs are explicitly designing cyclic peptides with oral bioavailability as a design constraint. D-amino acid substitutions — swapping the naturally occurring L-amino acid form for its mirror image — reduce protease sensitivity at specific positions, because most proteases evolved to cleave L-peptide bonds and are substantially less active against D-amino acid sequences. PEGylation and lipidation both affect oral bioavailability as well, though usually through effects on plasma half-life and distribution rather than GI permeation directly.

The commercial implications of getting oral delivery right are difficult to overstate. The GLP-1 market — semaglutide, tirzepatide, and the pipeline behind them — is already among the largest in pharmaceutical history by revenue. Injectable GLP-1 drugs require cold storage, trained administration, and needle familiarity that creates friction at every step of the access chain. An oral GLP-1 that performed comparably would collapse most of that friction and reach populations where injection adherence is a genuine barrier. Several oral GLP-1 candidates beyond semaglutide are in late-stage clinical development as of 2025 — Eli Lilly's orforglipron is a small-molecule GLP-1 agonist in Phase III, not strictly a peptide but part of the oral incretin story; other oral peptide GLP-1 agonists are in earlier phases. The race is real and the stakes are large.

The honest accounting of oral peptide delivery includes acknowledging what the engineering costs. SNAC-formulated oral semaglutide requires a substantially higher total dose than the injectable form to achieve a therapeutically equivalent plasma concentration — because even one-percent bioavailability means ninety-nine percent of the drug is being discarded. This has manufacturing cost implications, drug interaction implications (particularly the food-timing requirement for Rybelsus), and dose precision implications. Lipid nanoparticle formulations are expensive to manufacture at scale. The dose required for the SOMA-style approach depends on how much the device can contain and how reliably it functions across different individuals. None of these are insurmountable problems, but they're not solved by declaring oral delivery a success and moving on.

The patient experience improvement, though, is genuine. For many people — those who travel frequently and face cold-chain problems with injectable drugs, those managing needle phobia, those whose adherence to injectable medications erodes over months of weekly injections, those in clinical settings where injection training and monitoring capacity is limited — an oral option that works at all represents a fundamentally different relationship with their medication. That improvement in the real-world experience of treatment has clinical relevance beyond the pharmacokinetics. Drugs that get taken work better than drugs that don't.

The next decade of oral peptide delivery will be shaped by which of these engineering approaches scale, which peptide targets prove accessible to oral formulation, and whether the clinical outcomes from oral and injectable versions of the same compound prove equivalent enough to support prescribing substitution. The engineering has moved from theoretical to demonstrated, at least for some compounds. The question is no longer whether an oral peptide can work — Rybelsus answered that — but which peptides become workable next, and at what cost.