A 2020 mouse study hit an oral bioavailability no approved oral peptide drug has matched in a human. Why lipid-based delivery — the third route promised at the end of this series — still hasn’t been tested in a person.
BY GEORGE FORREST · 11 AUGUST 2026 · TECHNOLOGY
By Ibumix · 7-minute read
A 2020 study out of Beijing gave diabetic mice an oral insulin formulation that reached 28.7% of the bioavailability of a subcutaneous injection [1]. Rybelsus manages 0.4–1% in people [4]. Mycapssa manages under 1%, and under 0.2% at its higher dose [5]. Nobody has given the Beijing formulation to a person.
The best number in this series belongs to the battle nobody has fought yet.
Two answers, one route not yet tried
The first two articles in this series each found a drug that had already won its battle. Semaglutide, as Rybelsus, avoids Battle One — surviving stomach acid — by borrowing a stomach-based microenvironment from an excipient called SNAC, and barely has to fight Battle Two at all. Octreotide, as Mycapssa, takes Battle Two head-on, prising open the tight junctions between intestinal cells with sodium caprylate. Both are FDA-approved. Both work by performing a chemical or biological trick on the gut itself — changing, briefly, what the gut will allow through.
The third route promised at the end of the last article does neither. It changes what the drug travels in, not what the gut does.
A different kind of carrier
Lipid-based drug delivery systems — solid lipid nanoparticles, nanostructured lipid carriers, self-emulsifying systems — wrap the drug inside a fat-like particle before it ever reaches the stomach [3]. The idea is not to negotiate passage through the gut wall but to arrive already dressed as something the gut is built to absorb. A lipid particle can protect a fragile peptide from stomach acid and digestive enzymes on the way down, then get pulled into the body’s existing fat-absorption pathways rather than needing a bespoke chemical or biological workaround.
It is, in principle, a more general trick than either of the first two. SNAC’s microenvironment is tuned to specific molecules; tight-junction modulation has only been proven at the size of an eight-amino-acid peptide. A lipid carrier does not care as much what is inside it — which is exactly why the idea keeps attracting new attempts, and why so few of them have gone anywhere near a clinic.
Two attempts, two very different numbers
In 2023, a University of Innsbruck group working with the excipient maker Gattefossé took insulin glargine and made it lipophilic enough to load into a self-emulsifying drug delivery system (SEDDS), pairing it with sodium octadecyl sulfate before dispersing it in two lipid formulations built from commercially available excipients [2]. Dosed into rats by oral gavage, the two formulations reached absolute bioavailabilities of 0.55% and 0.44% — modest numbers, lower than Rybelsus’s — but a 7.7-fold and 6.2-fold increase over a plain aqueous solution of the same peptide [2]. The result is not dramatic. It is real, it is reproducible, and it is built from ingredients a formulator could buy today.
The Beijing study took a different approach entirely. Researchers at the Beijing Institute of Pharmacology and Toxicology encapsulated insulin inside a phospholipid shell, dissolved the whole particle in medium-chain triglyceride oil, and dosed it into healthy and diabetic mice, then diabetic rats [1]. Blood glucose in the diabetic mice dropped 70.2% within two hours and stayed suppressed for the rest of the study. In rats, the formulation reached 28.7% bioavailability relative to a subcutaneous injection of the same dose — over five times Rybelsus’s ceiling, and roughly thirty times what a single higher-content SEDDS formulation was managing in the Innsbruck study [1] [2].
The number and the drug are not the same thing
Nothing about that 28.7% has been tested in a person. Rodent and human gastrointestinal tracts differ enough — in transit time, bile physiology, and gut anatomy — that oral bioavailability rarely survives the jump from animal data to a clinical result unchanged. Rybelsus and Mycapssa both made that jump; both took years of clinical development to confirm, in people, numbers that started out looking promising in animals. A wider review of lipid-based carriers for peptide and protein drugs, published in 2024, is blunt about where the field still stands: plasma stability, membrane permeability and circulation half-life “remain a serious challenge which should be addressed in future” [3]. No lipid-encapsulated oral peptide has yet completed that journey to an approved medicine, the way SNAC and tight-junction modulation both have.
The best number in this series belongs to the battle nobody has fought yet — and until someone runs it in a person, that number is a promise, not a result.
Three battles, three answers, one still open
Stomach acid has an approved answer. The intestinal wall has an approved answer, at a cost. The third route — change the vehicle, not the gut — has the best numbers on paper and the least proof anywhere else. Three different technologies have each won one battle. None has had to win all three in the same molecule.
Ibumix is developing a series of new liquid drug delivery platforms built on glyceryl caprate (GCC) lipid chemistry, with a patent estate covering the underlying formulation and device.
Continue the series: Two Roads to One Pill looks at how Novo Nordisk and Eli Lilly each solved the same oral bioavailability problem on the market today — one tolerating it, one avoiding it.
Sources
- Wang T, Shen L, Zhang Y, Li H, Wang Y, Quan D. “Oil-soluble” reversed lipid nanoparticles for oral insulin delivery. J Nanobiotechnology. 2020;18(1):98. doi.org/10.1186/s12951-020-00657-8 — pmc.ncbi.nlm.nih.gov
- Claus V, Spleis H, Federer C, Zöller K, Wibel R, Laffleur F, Dumont C, Caisse P, Bernkop-Schnürch A. Self-emulsifying drug delivery systems (SEDDS): in vivo proof of concept for oral delivery of insulin glargine. Int J Pharm. 2023;639:122964. doi.org/10.1016/j.ijpharm.2023.122964 — pubmed.ncbi.nlm.nih.gov
- Mehrdadi S. Lipid-Based Nanoparticles as Oral Drug Delivery Systems: Overcoming Poor Gastrointestinal Absorption and Enhancing Bioavailability of Peptide and Protein Therapeutics. Adv Pharm Bull. 2024;14(1):48-66. doi.org/10.34172/apb.2024.016 — pmc.ncbi.nlm.nih.gov
- RYBELSUS (semaglutide) Full Prescribing Information — estimated absolute bioavailability of 0.4–1%. FDA label. accessdata.fda.gov
- MYCAPSSA (octreotide) Full Prescribing Information — absolute bioavailability below 1% (20 mg) and below 0.2% (80 mg) versus subcutaneous octreotide. FDA label. accessdata.fda.gov