Oral semaglutide is hard to deliver. Low membrane permeability and gut degradation keep unassisted bioavailability well under one percent. Rybelsus® solves that with SNAC. A 2026 Pharmaceutics paper (Kim et al.) tested sodium caprate (C10) as another enhancer in immediate-release tablets, and compared C10 with SNAC in Caco-2, rats, and dogs. Two things matter for a CDMO reader. First, C10 and a single-layer tablet got close to the SNAC reference in beagle dogs (relative Cmax 91.2%, relative AUClast 87.1%). Second, before any of that work, the authors screened three commercial semaglutide sources. Sinopep-Allsino’s lot sat in a 2–6 µm range they described as consistent and compact. That was the substance they took into dissolution, permeability, and PK. This page is a summary of that independent study. The C10 formulation is the authors’ work, not Sinopep technology. What the paper documents is the substance they chose, and why morphology was part of that choice.
The oral delivery problem, and why C10
SNAC works, but it needs high loading, fasting rules, and a stomach-limited, pH-dependent route. C10 is a medium-chain fatty acid with a food-additive history. It mainly opens tight junctions and changes epithelial membrane structure, rather than copying SNAC’s transcellular path. In Caco-2 cells, 10 mM C10 raised apparent semaglutide permeability 6.3-fold versus control, more than SNAC at twice the molar concentration. In rats, 300 mg/kg C10 raised AUClast about 33-fold versus peptide alone.
Semaglutide apparent permeability (Papp) fold-change vs.
control, Caco-2 monolayers
Figure 1. Semaglutide Papp fold-change vs. control, Caco-2, C10 vs SNAC.
Tablet design and dog PK
They built three immediate-release designs: single-layer blend (SSL-T), bilayer (SBL-T), and a compression-coated core/shell (SCC-T). SSL-T and SBL-T matched Rybelsus dissolution (f2 67.8 and 64.6). The coated tablet lagged (f2 36.2) because the shell delayed co-release. Rapid, co-localized release of peptide and enhancer was the better pattern. SSL-T went forward.
In dogs, a 14 mg C10 single-layer tablet vs the SNAC reference: relative Cmax 91.2%, relative AUClast 87.1%, half-lives 42.30 vs 42.32 h. No abnormal clinical findings in either group.
Beagle dog PK: C10-based tablet vs. SNAC-based
reference product
Figure 2. Relative Cmax and AUClast, C10 tablet vs SNAC reference, beagle dogs.
Why the substance lot mattered
Optical microscopy of three sources showed real spread. One lot: irregular 10–20 µm particles with agglomeration. Another: finer but less uniform, under 2 µm. Sinopep-Allsino: 2–6 µm, called consistent and compact. The authors tied that to blend uniformity and content homogeneity, then used that lot for every dissolution, Caco-2, and PK experiment in the paper.
Figure 3. Semaglutide particle size range by source (optical microscopy).
That finding has limits. It does not mean C10 is Sinopep IP, and it does not prove bioequivalence in humans. Caco-2 is not a full gut. Rat and dog GI physiology do not map one-to-one to people. Repeat-dose C10 safety was not in this single-dose study. The practical point is narrower: for oral peptide tablets, particle size and agglomeration are not cosmetic specs. They show up in blend uniformity and in how reproducible the downstream dissolution and absorption data are. The full white paper has the complete figures, the three tablet architectures, the dog PK numbers, and the sourcing discussion. To download the full white paper, click the button below.
