Oral Peptide Delivery: Why Almost Every Peptide Is an Injection

Delivery science

Oral peptide delivery is the hardest routine problem in peptide pharmacology: the gut is a machine purpose-built to take peptides apart, and it is extremely good at its job. Of the thousands of peptide drugs and candidates in existence, you can count the systemically absorbed oral products approved in the United States on one hand. Understanding why — and what the few successes actually did differently — explains more about peptide behaviour than almost any other topic.

Why oral peptide delivery fails in the gut

A peptide swallowed as a tablet runs a gauntlet with at least four independent kill mechanisms, any one of which is usually sufficient on its own.

Acid. The fasted stomach sits around pH 1–2. That alone hydrolyses acid-labile bonds and unfolds most structured peptides before any enzyme is involved.

Proteases. Pepsin works in the stomach; trypsin and chymotrypsin arrive from the pancreas in the duodenum; further peptidase activity waits at the intestinal brush border. This is not an unfortunate side effect — it is the digestive system doing exactly what it evolved to do to dietary protein, and a therapeutic peptide is chemically indistinguishable from dinner.

Mucus. The mucus layer over the epithelium is a dense, negatively charged, constantly renewed gel. Charged and hydrophobic peptides get trapped in it and cleared before they ever reach a cell membrane. Mucus thickness and composition vary along the gut and between species, which is one reason animal models overpredict human performance so consistently.

The epithelium itself. There are only two ways across. The paracellular route — between cells — is gated by tight junctions that are both size- and charge-selective, and a 4,000-dalton polar molecule is far too large for it. The transcellular route — through cells — requires membrane permeability that large polar peptides simply do not have, and anything that does get in faces efflux transporters pumping it back out.

What oral bioavailability actually looks like

These are label figures, not estimates. They are worth sitting with, because they reset expectations about what “oral” means for a peptide.

• Desmopressin tablets: roughly 0.16% of an oral dose reaches the circulation compared with intravenous administration.
• Oral semaglutide (Rybelsus), original formulation: absolute bioavailability approximately 0.4–1%.
• Oral semaglutide, newer formulation: approximately 1–2% — a genuine doubling achieved purely through formulation work.
• Oral octreotide (Mycapssa): the label does not state a percentage, but reports that a 20 mg oral dose produces exposure similar to a 0.1 mg subcutaneous dose — a 200-fold dose penalty.
• Cyclosporine oral solution: around 30%, and the outlier that proves the rule — it is a naturally cyclic, heavily N-methylated peptide, not a designed one.
• Linaclotide: plasma concentrations are below the limit of quantification at every approved dose. Pharmacokinetic parameters “cannot be calculated.”

That last entry is not a failure. Linaclotide is a 14-amino-acid guanylate cyclase-C agonist that acts on receptors on the luminal surface of the gut wall. It succeeds orally because it is never absorbed. Several peptide drugs work this way, and they are a useful reminder that “oral peptide” and “systemically absorbed peptide” are different problems.

How oral semaglutide gets through

The single most influential solution to oral peptide delivery is SNAC — salcaprozate sodium, a small absorption-enhancing molecule co-formulated into the tablet at a much larger mass than the drug itself. The label is unusually specific about where it works: absorption “predominantly occurs in the stomach,” with a time to peak concentration of about an hour.

SNAC does two things at once. It raises the local pH in a small pocket around the dissolving tablet, which shelters semaglutide from pepsin, and it promotes transport across the gastric epithelium by a transcellular route. Note what it does not do: unlike the tight-junction openers discussed below, SNAC is not primarily a paracellular agent. Published work also indicates it helps keep semaglutide monomeric rather than self-associated, which matters because monomers cross membranes and aggregates do not.

The consequences show up as dosing rules on the label, and they are strict:

• Take on an empty stomach, in the morning, with no more than 4 ounces of water.
• Wait at least 30 minutes before any food, drink or other oral medicine.
• Swallow the tablet whole — do not split, crush or chew.

Those rules are not caution for its own sake. Label pharmacokinetic studies found that a 10 mg dose taken with 50 mL of water produced higher exposure than the same dose with 240 mL, and that absorption rose with longer post-dose fasting across 15, 30, 60 and 120 minute intervals. Dilution and food genuinely move the number.

Clinically, the approach works. In the 26-week PIONEER 1 trial of 703 people with type 2 diabetes, placebo-adjusted HbA1c fell by 0.6, 0.9 and 1.1 percentage points at the 3, 7 and 14 mg doses respectively. At higher doses the effect scales: the 68-week OASIS 1 trial of 50 mg oral semaglutide in 667 adults with overweight or obesity reported mean bodyweight change of −15.1% versus −2.4% on placebo. In December 2025 the FDA approved a 25 mg once-daily oral semaglutide for weight management — the first oral GLP-1 receptor agonist cleared for that use.

The other approved route: oral octreotide

Mycapssa, approved in June 2020 for maintenance treatment of acromegaly in people already responding to injected somatostatin analogues, took a different path. It uses a transient permeability enhancer system built around sodium caprylate — a medium-chain fatty acid — suspended in caprylate glycerides inside a delayed-release capsule that opens in the intestine rather than the stomach.

The label’s food-effect data is the most striking number in this whole field: taking the capsule with food produced an approximate 90% decrease in both peak concentration and total exposure. Proton pump inhibitors, H2 blockers and antacids also reduce bioavailability enough that the label warns the dose may need to be increased. When absorption depends on a narrow chemical window, everything that shifts that window matters.

The full toolbox, and why most of it plateaus

Beyond the two approved systems, the field has converged on a fairly small set of tactics:

• Permeation enhancers. Medium-chain fatty acids such as sodium caprate (C10) transiently open tight junctions. Clinical results are single-digit percentages with high variability.
• Protease inhibitors. Soybean trypsin inhibitor and aprotinin protect the payload but need large co-doses and raise their own safety questions. Oramed’s oral insulin candidate used this approach and its phase 3 programme was discontinued early for lack of efficacy.
• Enteric coating. Skips the stomach entirely — useful, but it only removes one of five barriers.
• Backbone chemistry. Cyclisation and N-methylation reduce hydrogen-bond donors and protease recognition. Cyclosporine, with seven N-methylated backbone amides, is the proof this can work — though even it is described on its own label as “incomplete and variable” and, on chronic dosing, “erratic.”
• Lipidation. Attaching a fatty chain aids membrane interaction and albumin binding. AstraZeneca’s bis-lipidated GLP-1 candidate MEDI7219 reached around 6% bioavailability in dogs.
• Mechanical devices. The most inventive answer is to stop trying to cross the epithelium chemically and simply inject through it from inside. The self-orienting SOMA capsule, published in Science in 2019, is a swallowable device that rights itself against the stomach lining and fires a needle made of compressed drug into the wall, reaching plasma levels comparable to a subcutaneous injection in animal models. A later liquid version delivered a monoclonal antibody in large animals. Neither has been tested in people.

The honest summary of the field is that most of these tools, applied to most peptides, rarely push bioavailability past the low single digits in large animals or humans.

Why 1% is sometimes enough

The counterintuitive part of oral peptide delivery is that a bioavailability of 1% is not automatically disqualifying. What matters is whether the drug can absorb the resulting noise.

Semaglutide has an elimination half-life of about a week. Day-to-day swings in how much crosses the stomach wall are smoothed out across a dosing interval measured in days. Desmopressin, with a half-life of a couple of hours, has no such buffer — and its label duly notes that pharmacodynamic response between subjects was highly variable and that individual dosing is therefore recommended.

Long half-life, not high absorption, is the property that makes a low-bioavailability oral peptide workable. That is why the successful oral peptides so far have been either extremely long-lived in circulation, extremely potent per milligram, or not absorbed at all. If you want the underlying concepts, our guides to peptide half-life, peptide pharmacokinetics and ADME, and half-life extension strategies cover the machinery in more detail.

The cost is paid in dose. To match what 0.1 mg of subcutaneous octreotide delivers, the oral product uses 20 mg. Manufacturing a peptide at two hundred times the injectable scale is a real constraint on which molecules are even worth attempting orally — and it is one reason synthesis route and scale shape the commercial decision as much as the pharmacology does.

Frequently asked questions

Why can’t peptides just be swallowed like other drugs?

Because the digestive system is specifically designed to break peptide bonds. Acid, pepsin, pancreatic proteases and brush-border enzymes each attack the molecule, and even an intact survivor faces an epithelium whose gaps are too small and whose membranes are too impermeable for a large polar molecule. Small-molecule drugs escape this because they are not made of the thing the gut digests — a difference explored further in peptides versus small-molecule drugs.

Does a smaller peptide absorb better orally?

Size helps, but it is not sufficient. Cyclosporine is about 1,200 daltons and reaches roughly 30% bioavailability — but it is also cyclic and heavily N-methylated, which removes the hydrogen-bond donors and protease recognition sites that would otherwise sink it. Backbone chemistry matters more than mass alone, which is why macrocyclisation is such an active design area.

Are oral and injected doses of the same peptide interchangeable?

No, and not even close. Oral octreotide uses a 20 mg dose to approximate a 0.1 mg injection. Even two oral formulations of the same molecule can be non-interchangeable: the Rybelsus label carries an explicit conversion table between its two tablet series because milligram counts do not map across them.

Why do oral peptide labels have such strict food and water rules?

Because absorption depends on a narrow chemical window that food and fluid volume disrupt. Oral octreotide loses roughly 90% of its exposure when taken with food; oral semaglutide absorbs measurably better with 50 mL of water than 240 mL. These are not conservative defaults — they are the conditions under which the product was shown to work.

References

Informational only — not medical advice · 21+. Nothing here is a recommendation to take, buy or stop any medicine. Discuss any treatment decision with a qualified healthcare professional.