The Endotoxin Line on a Peptide COA

ON THIS PAGE

1. Key findings
2. What an endotoxin unit is
3. The limit is a calculation
4. Four units, no common denominator
5. "Endotoxin free" is not a finding
6. The 26 that name it without measuring it
7. Why a research buyer should care about a small number
8. Limitations and open questions
9. Sources

One hundred and twenty-six supplier dossiers sit in this ledger. Forty-four of them have endotoxin in the record somewhere, counted today. Eighteen ever produced a number.

That gap is the story, and the smaller number is worse than it looks, because the eighteen do not agree on what they are counting. The figures arrive as EU per millilitre, EU per milligram, EU per vial, and once as EU per kilogram in a footnote under a finding stated in a different unit. A buyer comparing two certificates is usually comparing two scales.

What an endotoxin unit is

The endotoxin unit is defined by activity against a reference standard, not by mass. That is why it survives the switch between assay chemistries. USP General Chapter <85> carries the compendial bacterial endotoxins test, and FDA's own teaching material lists its three formats as gel clot, turbidimetric, and chromogenic. The reagent underneath has been changing. The European Pharmacopoeia published chapter 2.6.32, a bacterial endotoxins test using recombinant factor C, in Supplement 10.3 on 1 July 2020, describing it as an alternative to "the classic limulus amoebocyte lysate (LAL)-based methods"; that chapter took force on 1 January 2021. USP's Microbiology Expert Committee approved Chapter <86>, covering recombinant cascade and recombinant Factor C reagents, in a notice dated 26 July 2024, official May 2025. USP's notice is explicit that <86> sits beside <85> rather than replacing it, and that manufacturers using LAL "can continue to do so."

So a certificate naming LAL under USP <85> is naming a current method, not an obsolete one. What matters is whether it names a method at all.

The limit is a calculation

Here is the part almost no storefront explains. A compendial endotoxin limit is not a constant somebody looked up. It is K divided by M, where K is the threshold pyrogenic dose of endotoxin per kilogram of body weight and M is the maximum recommended bolus dose of product per kilogram of body weight. The word bolus is load bearing for anyone doing the arithmetic. FDA's Center for Drug Evaluation and Research teaching deck on the endotoxins specification prints the K values it attributes to USP <85>: 5 EU/kg per hour for any route other than intrathecal, 0.2 EU/kg per hour intrathecal, and 100 EU per square metre of body surface area per hour where a product is given that way. The same deck notes FDA's assessment office uses 70 kg as the standard adult weight for the arithmetic.

Two consequences follow, and both are load bearing for anyone reading a certificate. First, the limit moves with the product, which is why FDA's Questions and Answers guidance says the fixed limit table from its 1987 guidance "is out of date" and directs firms to "use the calculation methods provided in the USP or AAMI standards." Second, the calculation takes a dose as its input. Research-use material has no dose. A specification line on a research peptide certificate is therefore borrowed from a framework the material is not inside, and a "Pass" against an unstated specification carries no information at all.

One document in this ledger shows the seam directly. A laboratory certificate for a 10 mg retatrutide lot, recorded 2026-08-25, reports endotoxins as passing under 5.00 EU/mg, beneath a footnote that states the pass criteria in EU/kg terms, 5 EU/kg and 350 EU total for a 70 kg adult. The finding and the criterion are in different units on the same page. A second document from the same laboratory in the same catalog reports 124.00 EU/mg over that identical footnote.

Four units, no common denominator

The dossier record, read across all 126 suppliers, shows how far apart the figures sit. Each supplier document in this section is quoted from the dossier entry that recorded it, on the date given, and none of the five was refetched for this piece. Virtus Peptides, recorded 2026-08-25, published one lot at not more than 0.05 EU/mL by USP <85> and two others at 0.07 and 0.142 EU/mL. Luxara Labs, recorded 2026-08-25, published a separate endotoxin certificate reporting under 0.116 EU/mg against an acceptance limit of under 0.15 EU/mg, naming the kinetic chromogenic method under USP <85>, the instrument, the standard curve range, and a spike recovery figure. That is the most complete endotoxin document in this corpus. Arcane Peptides, recorded the same day, reported one compound at under 5 EU per vial against an under-80 EU per vial specification and another at under 0.0945 EU/mg against an under-0.5 EU/mg specification. Same vendor, same certificate template, two units and two unrelated ceilings.

Two patterns in the record turn the unit problem into a claims problem. Disguised Alpha's customer-facing certificate image, recorded 2026-08-25, printed an endotoxins limit of 5 EU/mL with a passing mark, while the issuing laboratory's own record gave the same measured 0.080 EU/mL against a limit of under 0.25 EU/mL and a status of Reported rather than Pass. The measured value clears both ceilings. The customer-facing image stated a ceiling twenty times looser than the laboratory's and converted a report-only line into a pass. And VeriPeptides' verification tool, read 2026-08-26, printed the fixed string "< 0.5 EU/mg (LAL)" as the endotoxin line for every product in a 24-item catalog, cycling laboratory names by array position. A constant is not a measurement.

"Endotoxin free" is not a finding

An assay reports against a detection limit. It does not report absence. Licensed Peptides' homepage carries a badge reading "Endotoxin-Free" and a quality standard reading "0% Endotoxin tolerance", both fetched 2026-09-05. Its own BPC-157 product page, fetched the same day, prints an Endotoxin (LAL) row at under 1.0 EU/mg with the phrase LPS-Free beside it. A figure under a less-than sign is a limit, not an absence, and the two claims sit on the same site. The figure has also moved: when this ledger read that product page on 2026-08-23 the transcribed certificate line reported endotoxin by LAL at under 5.00 EU/mg, and the string 5.00 EU/mg is not on the page today. Excalibur Peptides' certificate page stated, when the ledger read it on 2026-08-24, that every batch is independently tested to be "endotoxin free," while the certificate under that claim states in its own Testing Purpose that the analysis "does not provide information on particulate matter, microbial contamination or presence of endotoxins."

The adjacent confusion is sensitivity. Several certificates in this corpus print "Assay Sensitivity: 0.05 EU/mL" beside two passing replicates. That figure is what the method can see, not what the batch contains, and a certificate that prints it with a less-than sign in front turns a detection floor into a specification the lot met. Koi Peptides Canada's endotoxin companion document went the other way. Recorded 2026-08-26, it reported two duplicate Pass findings against a USP <85> guideline reference, with no numerical EU/mg figure on the document. That one cannot be rechecked and is left in the past tense for a reason: koipeptidescanada.com did not resolve on 2026-09-05, returning "DNS name does not exist", while three other supplier domains resolved in the same run.

Sterility does not settle the question either. FDA's teaching deck lists endotoxins as not inactivated by most sterilization methods, and the 2018 review "Endotoxins from a Pharmacopoeial Point of View" in Toxins states that gram-negative bacterial endotoxins "are thermostable in the presence of moist heat and are not significantly destroyed by conventional autoclaving processes." A sterile vial can carry endotoxin. This is the same point the earlier piece on what a COA does not tell you makes about scope, arriving from the chemistry side.

The 26 that name it without measuring it

The larger group is the one with no figure at all, and its pattern is a policy page describing testing that the certificates do not contain. Cellugenix Peptides' quality page sets out a five-part sequence including an endotoxin quantification assay; the retatrutide certificate actually published, recorded 2026-08-25, carried no endotoxin figure. Apollo Peptide Sciences' About page promised a certificate and an endotoxin report for every product; none of the four laboratory reports read for that dossier on 2026-08-25 contained an endotoxin line. Evolve Peptides attached an endotoxin document to 33 of 81 batches when its batch lookup was read on 2026-08-23, a count that cannot be re-derived from the storefront because the lookup answers only to a batch number and enumerates nothing. Chameleon Peptides' testing panel, recorded 2026-08-24, listed endotoxins with purity, mass, identity, heavy metals and sterility as "Pending, Final report not yet published".

Two suppliers sit inside the group of 18 only because the number in their record is a criterion or a specification rather than a finding, which is the same gap wearing a number. Grey Research Peptides' quality page, recorded 2026-08-25, described endotoxin testing by the Limulus Amebocyte Lysate method against a stated acceptance criterion of less than 1 EU/mg, in the same list as the HPLC and mass-spectrometry work it calls standard; none of the 32 certificates obtained from the laboratory's own domain carried an endotoxin line, and the same page's FAQ described endotoxin testing as something a buyer must separately request. Forward Peptides' certificate, recorded 2026-08-25, listed Endotoxin (LAL) against a specification of less than 5.0 EU/mL and reported it PENDING, so the ceiling was on the page and the finding was not. Counted by unit, both belong to the 18. Counted by what a buyer can actually read off a document, neither does.

Why a research buyer should care about a small number

Not for any reason involving a person. The stake is the experiment. Schwarz et al., writing in PLoS One in December 2014, note that suppliers of recombinant proteins commonly guarantee contamination under 1 EU per microgram, then report that in their in vitro work "even 0.02 ng/ml LPS was able to significantly stimulate the production of cytokines", and conclude that when working with cells particularly sensitive to LPS, "even low endotoxin contamination may generate erroneous data." That quoted sentence covers all four preparations stimulated in that figure, which the paper names as the monocytic cell line THP-1, primary human monocytes, in vitro differentiated human monocyte-derived dendritic cells and primary human CD1c+ dendritic cells, read at the journal on 2026-09-05. Only the last of those carries the title finding. It is an in vitro finding and it is enough to make the point: below the threshold a vendor advertises is not the same as below the threshold an assay tolerates. Which is exactly what a bare "Pass" refuses to tell you.

Limitations and open questions

These counts come from a text search of dossier notes, not from a certificate-by-certificate audit. The counting rule, stated so it can be checked: a dossier joins the 18 when its notes quote a number standing in endotoxin units, a compendial chapter number such as USP <85> does not count as one, and a note recording that a document carries no EU/mg figure counts as an absence rather than as a figure, which is why Koi Peptides Canada sits in the 26 and not the 18. A supplier whose documents carry an endotoxin line that no dossier note happened to quote falls outside the 18 as well. Every named-supplier observation here describes what a vendor page or a laboratory document showed on the date recorded in its dossier entry, and storefronts change: those entries run from 2026-08-23 to 2026-08-26, while the ledger as a whole holds dated entries from 2026-08-17 to 2026-09-05. Of the supplier observations, only Licensed Peptides' pages and the Koi Peptides Canada domain were refetched on 2026-09-05: the first had changed its endotoxin figure, the second no longer resolved. The absence of an endotoxin document from a storefront is not proof that no assay was run. USP <85> and <86> sit behind USP's subscription and were not read directly; the K values above are quoted as FDA restates them, and any reader with access should check them at source. Most of all, the compendial framework described here governs drug products and medical devices. Research-use material is outside it, which is the reason a limit line on one of these certificates needs a stated basis before it means anything.

The unit is the cheapest thing on the page to fix, and the one most often left off.

Sources

Primary regulatory documents, all read 2026-09-05: FDA, Guidance for Industry, Pyrogen and Endotoxins Testing: Questions and Answers (Edition 2), March 2026, issued by CDER, CBER, CVM, CDRH and ORA, for the statements on the 1987 limit table, the USP and AAMI calculation, and device limits in EU/mL and EU per device; FDA CDER, The Bacterial Endotoxins Specification: Points to Consider, a Small Business and Industry Assistance presentation by Erika Pfeiler of the Office of Pharmaceutical Manufacturing Assessment, for the K/M calculation, the K values attributed to USP <85>, the 70 kg standard weight, and the statement that endotoxins are not inactivated by most sterilization methods. That deck carries no publication date on its face; its own resource list cites a March 2024 guidance, so it is no earlier than that. FDA, FDA Clarifies Current Thinking on Pyrogen and Endotoxins Testing, updated 18 March 2026, for the Edition 2 revision and recombinant reagents.

Standards bodies, independent of FDA and of each other: USP, Expert Committee approves endotoxin testing using non-animal derived reagents, dated 26 July 2024, read 2026-09-05, for Chapter <86>, its November 2024 early adoption and May 2025 official date, and its relationship to <85>. The European Directorate for the Quality of Medicines and HealthCare, "Recombinant factor C: new Ph. Eur. chapter available as of 1 July 2020," Strasbourg, dated 01/07/2020, read 2026-09-05, for chapter 2.6.32, its publication in Supplement 10.3, the revised chapter 5.1.10, and the 1 January 2021 date it took force.

Peer-reviewed literature, both read 2026-09-05: Franco E, Garcia-Recio V, Jimenez P, Garrosa M, Girbes T, Cordoba-Diaz M, Cordoba-Diaz D, "Endotoxins from a Pharmacopoeial Point of View," Toxins (Basel) 2018;10(8):331, for endotoxin thermostability under moist heat. Schwarz H, Schmittner M, Duschl A, Horejs-Hoeck J, "Residual Endotoxin Contaminations in Recombinant Proteins Are Sufficient to Activate Human CD1c+ Dendritic Cells," PLoS One 2014;9(12):e113840, for supplier guarantee levels and the in vitro activation figures.

Supplier observations are this publication's own, from the dossier record in ledger/vendors.json. Each dossier entry carries the date it was recorded. The entries behind the named-supplier observations here run from 2026-08-23 to 2026-08-26; the ledger as a whole holds 607 dated entries running from 2026-08-17 to 2026-09-05, and the two ranges are different things. Counts in the key findings were derived from that file on 2026-09-05 rather than carried from an earlier read. The two vendor pages refetched on 2026-09-05 are named beside the observations that rest on them. Vendors are named, never linked, per site policy.