Decapeptide-52 (OS-01)_Dosage, benefits & legal status

The claim here is unusual for a cosmetic peptide. Most are sold for what they do to the look of skin. This one is sold for what it is said to do to skin's biological age, and that requires understanding what was measured and in what.

1. 01
   Senescent cells, and the word that matters

   As tissue ages, some cells stop dividing but refuse to die, and they leak inflammatory signals that degrade the tissue around them. Two strategies exist against them. A senolytic kills them. A senomorphic leaves them alive but quiets what they secrete. The developer's own peer-reviewed papers describe this peptide as senomorphic. That distinction is the hinge of the one safety point on this page, because the oral product on sale markets it as a senolytic, contradicting the published characterization of the molecule it claims to contain.
2. 02
   What the epigenetic clock actually is

   DNA carries chemical tags at specific positions, and the pattern of those tags shifts predictably with age. A clock is a statistical model trained to read age from that pattern. The one used here was built on 2,266 positions from 249 training and 259 test samples, with the test samples drawn from the same original dataset, so the test set is not independent, plus an external check on 16 whole skin biopsies giving a correlation of 0.95 and a typical error of 4.98 years. The important limitation is in what it was trained for: predicting chronological age. Doing that well does not establish that it can detect age changes caused by an intervention.
3. 03
   The 2.6 years, read closely

   Methylation age fell by an average of 2.6 years with p below 0.01, in skin taken from one 79-year-old female donor and kept alive outside the body. The rapamycin comparator in the same experiment was not significant at p=0.21. Two details inside the paper deserve attention. Its Methods report three replicates per condition while its Figure 6 legend reports four biological replicates, both in the same published article. Separately, the comparison the 2.6-year figure is measured against shifts between the running text, where it is framed against rapamycin, and Figure 6i, where it is against untreated controls.
4. 04
   Our own arithmetic, offered as arithmetic

   The claimed effect of 2.6 years is a little over half the clock's own external validation error of 4.98 years. Neither paper makes that comparison, so it is our observation rather than a published criticism. It does not make the result wrong: a mean shift smaller than a single-measurement error can be real and statistically detectable across replicates. It does mean the headline number should not be read as something measurable on an individual person's skin.
5. 05
   The sequence problem

   Across all of the peer-reviewed literature, the molecule is named and never spelled out. None of the papers says the sequence is being withheld; it simply is not given. A reader of the science therefore cannot learn what the compound is. The ten residues appear in one place, a table in a granted patent, as the first listed sequence identifier. Our own provenance caveat: the public patent viewer does not render that table, so the link between the name Peptide 14 and that sequence identifier came from a third-party patent reader, while the sequences themselves are confirmed in the patent document.