Cartalax (AED): what the published research actually shows

Has Cartalax been tested in people?

The finding, first: no study of the AED tripeptide in human subjects appeared in anything we searched. There is one sentence in one review that reads as though it says otherwise, and it is worth taking apart slowly, because everything downstream of this compound rests on it.

The sentence: a 2023 Russian-language review in Advances in Gerontology, PMID 37782637, credits two things with high efficacy in animal models of osteoarthritis and in oral use by older patients with the disease. One is Sigumir, a polypeptide complex of cartilage and bone tissue from young animals. The other is the AED tripeptide, which that review spells Kartalax, and which is the only place in the indexed literature where the trade name appears in an abstract at all.

What the English literature says about the same work: the polypeptide complex isolated from animal cartilage tissue is the substance in the second phase of clinical trials for osteoarthritis in Russia, and AED is one of the short peptides inside that complex, which runs from 75 to 846 Da (Linkova, 2023).

So the human trial belongs to a mixture, not to the molecule. A tripeptide that is one constituent of a multi-component preparation does not inherit that preparation's clinical record, any more than one amino acid inherits the record of a protein it sits in.

The second reason to hold that review at arm's length: it is not open access, Europe PMC records it as having no retrievable reference list, and its references endpoint returned zero entries when we called it on 4 August 2026. The patient sentence inside it cannot be traced to any underlying study, by us or by anyone else reading from outside that institute.

Every published experiment, and what each one actually used

Why the whole list fits on one screen: this is not a summary of the evidence. It is the evidence, all of it that we could retrieve, one row per paper. A page about a compound with a thin file should show the file rather than describe it.

The detail the table is built to make unmissable: five of the eight used human cells, and human cells in a dish are not people. Three of the six concept-indexed records carry the PubMed Humans tag, and all three of those are cultures: a thymocyte culture, periodontal ligament stem cells, and an embryonic bone-marrow line. Row 7 names no species in its English abstract and carries the same tag with no Animals tag beside it, which is the only reason we count it as human cells. The tag describes where the cells came from, not who the subjects were.

The one that most looks like a human result and is not: the 2016 paper is titled for skin fibroblasts with no species in the title, and its indexing terms are Animals and Rats, Wistar. The 2023 English review attributes the same result to a model of replicative aging of rat skin fibroblasts, which is the second, independent confirmation that the cells were rat.

One more thing the table cannot show: every one of these papers carries an author from the Saint Petersburg Institute of Bioregulation and Gerontology, the institute where the compound was developed. That is not a reason to discard the work. It is a reason not to read eight papers as eight independent replications. The same pattern is worth checking on any compound whose file is this small, which is the habit the published record for PE 22-28 was written to build.

Why does searching the name Cartalax find almost nothing?

Because the name is a commercial label, not a term the literature uses. On 4 August 2026, PubMed searched for cartalax[tiab] returns a count of 0, and cartalax[ti] returns 0. Searched for kartalax[tiab], it returns 1, which is the review above. The name is not absent from the literature altogether, it simply does not live in the fields those searches read: the first row of Table 5 in a 2021 systematic review in Molecules, PMID 34834147, pairs the sequence AED with the name Cartalax.

Why a plain search still seems to work: PubMed searched for cartalax across all fields, run 4 August 2026, returns 6 records, and its own query translation shows why. It rewrites the word to "alanyl glutamyl aspartic acid"[Supplementary Concept] OR "alanyl glutamyl aspartic acid"[All Fields]. The database is quietly answering a question about the sequence while the reader thinks it answered one about the name.

The control that makes those zeros mean something: the same instrument on the same day returns 62 for epitalon[tiab], 15 for pinealon[tiab] and 81 for thymogen[tiab]. The zeros are real zeros, not a broken search.

The same shape in the other index: Crossref, queried on titles on 4 August 2026, returns 0 results for cartalax and 0 for kartalax, against 21 for epitalon, 28 for thymogen, 6 for pinealon and 1 for vesugen. Sibling compounds in this family have title-level literature under their trade names. This one does not.

What follows for anyone checking a claim about it: the productive search term is the sequence or the chemical name, not the label. A name that returns nothing in two independent indexes is not evidence of a conspiracy or of suppression. It is evidence that the name was coined for a product rather than for a paper.

Is the sequence Ala-Glu-Asp or Asp-Glu-Asp?

It is Ala-Glu-Asp, and the two databases that define this compound both print the wrong answer beside the right one. The MeSH supplementary concept record lists the entry terms Ala-Glu-Asp, H-Asp-Glu-Asp-OH, cartalax and T-31 peptide. The PubChem synonym list for CID 87815447 carries both H-Ala-Glu-Asp-OH and H-Asp-Glu-Asp-OH.

Both strings cannot describe one molecule, because alanine and aspartic acid are different residues and swapping them at position one changes the formula.

What settles it, in the same record: PubChem's computed IUPAC name for that CID reads (2S)-2-[[(2S)-2-[[(2S)-2-aminopropanoyl]amino]-4-carboxybutanoyl]amino]butanedioic acid. The N-terminal residue is an aminopropanoyl group, which is alanine. Aspartic acid at that position would carry a second carboxyl on its side chain and the name would say so.

The same answer from the connectivity string: the record's SMILES opens CC(C(=O)N, a methyl carbon on the first residue. That is alanine's side chain. The molecular formula C12H19N3O8 and mass 333.29 belong to the alanine form.

And the patent agrees: RU2299741 claims the peptide alanyl-glutamyl-aspartic acid of the formula H-Ala-Glu-Asp-OH, SEQ ID NO:1. Three independent records say alanine at position one; one synonym string says aspartate, and it is the only thing that does.

The transferable habit: a synonym list is a list of things people have called something, and a structure field is derived from the deposited structure. When the two disagree, the structure is the one to read. Two labels for one molecule is a recurring failure in this whole category, and it is the same trap as the two numbers attached to follistatin. The same question asked of a vial rather than of a database is what a certificate of analysis exists to answer.

What does the patent actually claim?

The document: RU2299741, filed by Khavinson, Grigoriev, Malinin and Ryzhak, held in the Europe PMC patent index with a first publication date of 30 May 2006. It is the primary document that defines what this compound is for, and it is worth being precise that a patent is an intellectual property grant rather than a marketing authorization.

The three things it proposes, in the patent's own terms: a pharmaceutical composition containing the peptide alanyl-glutamyl-aspartic acid as active component; the peptide itself, described as normalizing metabolism in bone and cartilage tissue; and a method for prophylaxis and treatment of the locomotor system, in particular degenerative-dystrophic joint and backbone disease.

The amount written into the method claim: a range of 0.01 to 100 mcg per kg of body mass, at least once daily, for the time needed to reach a therapeutic effect. That is a span of four orders of magnitude, a ten-thousand-fold band from the bottom of the range to the top.

Why the range is that wide, and what it tells you: a patent claim is drafted to cover ground, not to report a measurement. A tested amount narrows as evidence accumulates; a claimed amount widens to keep competitors out. The record's own bibliographic tail reads 6 cl, 2 tbl, 1 dwg, 1 ex, which is six claims, two tables, one drawing and one worked example behind a band that wide.

So that figure appears here and nowhere else on this site. It is not in any chart, it is not a default in any tool, and it is not a recommendation. It is the width of a legal claim, reported because the width is the informative part.

What is the minus 26.65 score, and what is it not?

Where it comes from: a 2022 review in the International Journal of Molecular Sciences prints a table of ultrashort peptides docked against LAT1, the human L-type neutral amino-acid transporter. The row reads Cartalax (AED), ICM-Score minus 26.65, biological activity chondroprotector. It is the only Cartalax-specific number in the peer-reviewed English literature that we found.

How it was produced: by computer docking of conformationally flexible compounds using ICM-Pro from Molsoft, against published spatial structures of LAT1. Nothing was measured in a cell, in an animal or in a person to generate it.

The review's own calibration: the known LAT1 ligands in its previous table score between minus 19.67 and minus 15.00, and it states that peptide scores below minus 15 may indicate a high probability of efficient binding. On that scale minus 26.65 sits well inside the range, and the same review reports that 20 of the 27 peptides in its table score no worse than those native ligands.

A discrepancy we found in the source and are not going to smooth over: the caption of that table names LAT1, while the sentence directly beneath it describes the same table as docking results in the NSP5 active site. Two different targets are named for one set of numbers. We report the figure as the table prints it and flag the inconsistency rather than pick the reading that sounds better.

What a docking score is not: a measured binding affinity, an effect in a cell, or a reason to expect anything in a body. It is a prediction about whether one molecule fits a pocket in a model, and its usefulness is in ranking candidates for an experiment somebody still has to run.

Is Cartalax approved as a medicine anywhere?

What we checked and what came back, each with a positive control so a zero cannot be a broken query: the openFDA drug label endpoint returned NOT_FOUND for cartalax and for kartalax, while the same endpoint returned 18 labels for semaglutide. DailyMed returned 0 records for cartalax and 9 for semaglutide. The trials registry returned 0 for kartalax and one fuzzy match for cartalax, which on inspection is a trial of an enteral nutrition formula and not a study of this compound.

The contrast that is specific to this family: on 16 April 2026 the FDA published notice of a Pharmacy Compounding Advisory Committee meeting on substances nominated for the section 503A bulk drug substances list. The notice names seven: BPC-157, KPV, TB-500, MOTs-C, Emideltide, Semax and Epitalon. Epitalon is the closest sibling to this compound in the same peptide family, and it reached a United States advisory committee agenda. Cartalax and Kartalax appear zero times in that document.

What we could not establish, said plainly rather than converted into an absence: the European Medicines Agency search endpoint returned HTTP 403 for our query and for the control query alike on 4 August 2026, so that is a blocked instrument and not a zero. The Russian state drug register responded, but it is a form we could not query programmatically, so registration status in Russia is unverified rather than absent.

The one thing that must not be read as approval: the patent. A granted patent means an office judged a claim novel. It says nothing about whether a regulator has reviewed safety or efficacy, and the difference between the two is the whole substance of what an actual approval covers.

What the record does not settle

Four open items, and none of them closes with a better search:

• Whether anything happens in a person. Eight cell and tissue experiments, one patent and one docking score do not add up to a human result, and no amount of adding them changes that.
• Whether the results replicate outside one institute. Every retrieved paper shares an affiliation with the group that developed the compound, so the file is deep in one place and empty everywhere else.
• What the underlying Russian data says. Four of the cited papers are Russian-language, not open access, and carry no retrievable reference lists, so their claims terminate at the abstract.
• What is in any particular vial. Nothing in this literature describes a product. Identity and purity are questions for a certificate, and even then a purity number leaves several questions open.

The honest summary of the file: this is a compound with a defined structure, a named inventor group, a patent describing an intended use, and no evidence in humans that we could locate. Each of those is a real fact and none of them substitutes for the missing one.

How this page is sourced

What was retrieved directly: every PubMed abstract listed below was read in full, the MeSH and PubChem records were pulled from their own endpoints, the patent claims were read from the Europe PMC patent record, and the full texts of the two 2022 and 2023 English reviews were downloaded and read at the table level rather than the abstract level. The Federal Register notice was read from the government printing office copy.

What was verified rather than assumed: every DOI below was resolved against Crossref and matched on journal, year, volume and pages, and every PubMed identifier was fetched and matched against the citation printed here. The species attribution on the 2016 fibroblast paper was checked against its own indexing terms, not inferred from its title.

What was deliberately left out: one further paper comparing AED and KE in human skin fibroblasts exists and its metadata checks out, but it sits behind a paywall and we did not read it, so no figure from it appears anywhere above.

Sixteen sources: eight primary experiments, four reviews, two reference-database records, one patent and one regulator document.

The standard: the full sourcing and citation-verification standard for this site, including who checks a page before it ships, lives on the methodology page.

Last reviewed: 4 August 2026.

1. 1

   NCBI MeSH Supplementary Concept Record, alanyl-glutamyl-aspartic acid, UID 67572340. Entry terms Ala-Glu-Asp, H-Asp-Glu-Asp-OH, cartalax and T-31 peptide; scope note, a synthetic peptide with geroprotective activity. Retrieved 4 August 2026. MeSH UID 67572340.

2. 2

   PubChem Compound Summary CID 87815447, alanyl-glutamyl-aspartic acid. C12H19N3O8, molecular weight 333.29, InChIKey KXEVYGKATAMXJJ-ACZMJKKPSA-N, CAS 85806-95-7, ChEBI:158137; synonym list includes Cartalax, T-31 peptide, H-Ala-Glu-Asp-OH and H-Asp-Glu-Asp-OH. Retrieved 4 August 2026. PubChem CID 87815447.

3. 3

   Khavinson VKh, Grigoriev EI, Malinin VV, Ryzhak GA. Peptide normalizing metabolism in osseous and cartilage tissue, pharmaceutical composition based on thereof and method for its using. Patent RU2299741. Europe PMC patent record, first publication date 30 May 2006; full English abstract read 4 August 2026. europepmc.org, RU2299741.

4. 4

   Lin'kova NS, Polyakova VO, Trofimov AV, Kvetnoy IM, Khavinson VKh. Peptidergic regulation of thymocyte differentiation, proliferation, and apoptosis during aging of the thymus. Bull Exp Biol Med. 2011;151(2):239-242. doi:10.1007/s10517-011-1298-8. PMID 22238759.

5. 5

   Khavinson VKh, Tarnovskaia SI, Lin'kova NS, et al. [Tripeptides slow down aging process in renal cell culture]. Adv Gerontol. 2014;27(4):651-656. Russian language, English abstract read 4 August 2026; no DOI exists for this record. PMID 25946838.

6. 6

   Chalisova NI, Lin'kova NS, Nichik TE, Ryzhak AP, Dudkov AV, Ryzhak GA. Peptide regulation of cells renewal processes in kidney tissue cultures from young and old animals. Bull Exp Biol Med. 2015;159(1):124-127. doi:10.1007/s10517-015-2906-9. PMID 26033601.

7. 7

   Lin'kova NS, Drobintseva AO, Orlova OA, Kuznetsova EP, Polyakova VO, Kvetnoy IM, Khavinson VKh. Peptide regulation of skin fibroblast functions during their aging in vitro. Bull Exp Biol Med. 2016;161(1):175-178. Indexing terms include Animals and Rats, Wistar. doi:10.1007/s10517-016-3370-x. PMID 27259496.

8. 8

   Caputi S, Trubiani O, Sinjari B, Trofimova S, Diomede F, Linkova N, Diatlova A, Khavinson V. Effect of short peptides on neuronal differentiation of stem cells. Int J Immunopathol Pharmacol. 2019;33:2058738419828613. doi:10.1177/2058738419828613. PMID 30791821.

9. 9

   Ashapkin V, Khavinson V, Shilovsky G, Linkova N, Vanuyshin B. Gene expression in human mesenchymal stem cell aging cultures: modulation by short peptides. Mol Biol Rep. 2020;47(6):4323-4329. doi:10.1007/s11033-020-05506-3. PMID 32399807.

10. 10

    Khavinson VK, Popovich IG, Linkova NS, Mironova ES, Ilina AR. Peptide regulation of gene expression: a systematic review. Molecules. 2021;26(22):7053. Table 5 row 1 pairs the names AED and Cartalax. doi:10.3390/molecules26227053. PMID 34834147.

11. 11

    Khavinson V, Linkova N, Kozhevnikova E, Dyatlova A, Petukhov M. Transport of biologically active ultrashort peptides using POT and LAT carriers. Int J Mol Sci. 2022;23(14):7733. Table 2, LAT1 docking, Cartalax (AED) ICM-Score minus 26.65; full text read for the table and the ICM-Pro method sentence. doi:10.3390/ijms23147733. PMID 35887081.

12. 12

    Myakisheva SN, Linkova NS, Kozhevnikova EO, Polyakova VO, Ryzhak GA. [Peptides prevent the forming of secretory phenotype of chondrocytes associated with the aging]. Adv Gerontol. 2023;36(2):234-238. Russian language, English abstract read 4 August 2026; no DOI exists for this record. PMID 37356100.

13. 13

    Linkova N, Khavinson V, Diatlova A, Myakisheva S, Ryzhak G. Peptide regulation of chondrogenic stem cell differentiation. Int J Mol Sci. 2023;24(9):8415. Full text read for the polypeptide complex passage and the rat attribution. doi:10.3390/ijms24098415. PMID 37176122.

14. 14

    Myakisheva SN, Linkova NS, Kozhevnikova EO, Ryzhak GA. [Chondrocytes secretory phenotype associated with aging: role in the pathogenesis of osteoarthritis and prospects for peptide bioregulation]. Adv Gerontol. 2023;36(3):313-323. Russian language, English abstract read 4 August 2026; no DOI, not open access, and its Europe PMC references endpoint returns zero entries. PMID 37782637.

15. 15

    Myakisheva SN, Linkova NS, Diatlova AS, Polyakova VO, Ryzhak GA. [The influence of peptides on the chondrogenic differentiation of human mesenchymal stem cells during replicative aging]. Adv Gerontol. 2023;36(3):383-390. Russian language, English abstract read 4 August 2026; no DOI exists for this record. PMID 37782646.

16. 16

    U.S. Food and Drug Administration. Pharmacy Compounding Advisory Committee; Notice of Meeting; Establishment of a Public Docket; Request for Comments, Bulk Drug Substances Nominated for Inclusion on the Section 503A Bulk Drug Substances List. 91 FR 20465, 16 April 2026, Docket FDA-2025-N-6895, FR Doc 2026-07361. Retrieved and read 4 August 2026. govinfo.gov, FR Doc 2026-07361.

Related pages

Nearby on the same problem: five pages about reading a thin file, a confusing name, or a piece of paper that comes with a vial.

• PE 22-28, the published record: another compound whose file is small enough to read end to end, and what that changes.
• Follistatin 344 against 315: one protein, two numbers, and how a naming split turns into a purchasing mistake.
• What an approval actually covers: the difference between a granted patent, a marketing authorization and a label.
• How to read a COA: where identity, purity and content sit on a certificate, and what each one measures.
• What a purity number leaves out: why a high percentage on a chromatogram answers a narrower question than it looks like it answers.

The compound files: each new one is read and published first in The Decadewise briefing.