Bacterial Endotoxin Testing

Endotoxin is lipopolysaccharide — a structural component of the outer membrane of Gram-negative bacteria. Its pyrogenic activity resides almost entirely in lipid A, the acylated glucosamine disaccharide recognized in mammals by TLR4/MD-2.

Three properties define the analytical problem, and each of them defeats an intuition that is common in this market:

• It is not alive. Endotoxin is a molecule. Killing the organism does not remove it; lysing the organism releases it. A rigorously sterile material can carry a very high endotoxin burden.
• It is thermostable. Standard moist-heat sterilization at 121 °C for 15 minutes does not meaningfully degrade LPS. Depyrogenation requires dry heat — typically ≥250 °C with a validated ≥3-log reduction against an endotoxin challenge, the approach described in USP ⟨1228.1⟩ — or a chemical or chromatographic removal step.
• It passes sterilizing filters. LPS monomers are roughly 10–20 kDa and aggregate into micelles and vesicles. A 0.22 µm filter removes essentially none of it. Reduction requires ultrafiltration at 10–30 kDa MWCO, anion exchange, or affinity capture.

For research materials — particularly lyophilized peptides handled in facilities with no segregation controls — endotoxin is an independent quality attribute that no purity, identity, or content test touches.

The LAL cascade. Limulus amoebocyte lysate contains a serine-protease zymogen cascade: endotoxin activates Factor C, which activates Factor B, which activates the proclotting enzyme, which cleaves coagulogen to coagulin and forms a gel. A parallel branch matters analytically — Factor G is activated by (1→3)-β-D-glucans, not by endotoxin, and feeds into the same proclotting enzyme. Cellulosic and depth filters, glucan-containing excipients, and some fermentation-derived materials therefore produce false-positive LAL responses. This is managed with glucan-blocking buffers or by moving to a Factor C-only reagent.

USP ⟨85⟩ techniques, harmonized with Ph. Eur. 2.6.14 and JP 4.01, comprise three techniques in two variants each:

• Gel-clot, limit test — firm gel on 180° inversion, pass/fail against a limit at the lysate's labeled sensitivity λ (commonly 0.03–0.25 EU/mL). This is the referee method in a dispute between techniques.
• Gel-clot, semi-quantitative — endpoint of a two-fold dilution series; resolution is one two-fold step.
• Kinetic turbidimetric — time to a threshold optical density, or rate of OD change; wide dynamic range, tolerant of some colored matrices.
• Kinetic chromogenic — time to threshold absorbance at 405 nm from p-nitroaniline release; the most sensitive of the LAL formats and the usual quantitative choice.
• Endpoint chromogenic and endpoint turbidimetric — absorbance or turbidity after fixed incubation; narrower range, less used.

Kinetic methods rely on a log-log relationship between endotoxin concentration and onset time. USP ⟨85⟩ requires the absolute value of the standard curve correlation coefficient to be ≥ 0.980.

Recombinant reagents. Recombinant Factor C (rFC) uses Factor C alone with a synthetic fluorogenic substrate; recombinant cascade reagent (rCR) reconstitutes Factor C, Factor B, and the proclotting enzyme recombinantly. Because there is no Factor G, rFC is immune to β-glucan false positives — a genuine specificity advantage, not only a supply-chain one — and lot-to-lot consistency is generally better than animal-derived lysate. USP ⟨86⟩, covering endotoxin testing with recombinant reagents, was published for early adoption in November 2024 with an official date of 1 May 2025; USP has characterized the ⟨86⟩ methods as alternatives to ⟨85⟩ under General Notices 6.30, so for a compendial article their use still carries the alternative-procedure burden of demonstrating equivalence. Ph. Eur. has had chapter 2.6.32 for rFC since 2021.

PLACEHOLDER: which BET method(s) Arcadia actually performs — gel-clot, kinetic chromogenic, kinetic turbidimetric, rFC/rCR — and the reader or incubator used. Delete any method not actually offered

The Endotoxin Unit is a bioactivity unit, defined by activity against the USP Endotoxin Reference Standard, not by mass. As a rough conversion 1 EU corresponds to roughly 0.1 ng of the E. coli reference standard endotoxin, but that ratio is specific to that LPS chemotype — a different organism's LPS at the same mass can have very different activity. EU is never presented as a mass unit. Working reagents are Control Standard Endotoxin (CSE), which must carry a potency assigned against the RSE for the specific lysate lot in use; a result citing an EU value without identifying the CSE lot and its RSE-referenced potency is incomplete.

Two numbers govern every run. λ is the labeled sensitivity of the lysate (gel-clot) or the lowest concentration on the standard curve (photometric). MVD is the greatest dilution at which the endotoxin limit can still be detected:

> MVD = (endotoxin limit × concentration of sample solution) ÷ λ

With a limit of 0.5 EU/mg, a sample prepared at 10 mg/mL, and λ = 0.005 EU/mL, MVD = 1000-fold. Any dilution up to 1:1000 is valid; beyond that the test cannot see the limit. The report states the prepared concentration, the dilution tested, and the calculated MVD, with confirmation that the tested dilution did not exceed it.

Interference screening is not optional. Peptides, surfactants, chelators, extreme pH, and high protein content all perturb the enzymatic cascade. USP ⟨85⟩ requires an inhibition/enhancement determination on each new article:

• Gel-clot — the geometric mean endpoint of the spiked sample must fall between 0.5λ and 2λ.
• Photometric — recovery of the positive product control spike, after subtracting endogenous endotoxin, must fall within 50%–200%.

Additional validity conditions: the negative control (water for BET) must not exceed the reagent's blank limit; the standard curve must give |r| ≥ 0.980; and the pH of the lysate/sample mixture must sit inside the reagent manufacturer's window, usually 6.0–8.0. Where recovery fails, the permitted remedies are dilution to a lower concentration still below MVD, a more sensitive lysate, or a validated treatment — and any treatment must itself be shown to remove interference without removing endotoxin, demonstrated by spiking before treatment.

Every reported result carries its PPC recovery. A result with no positive product control is an uncontrolled result.

Low Endotoxin Recovery. Endotoxin spiked into certain matrices becomes progressively undetectable over hours to days. The classic trigger is a chelating buffer — citrate or phosphate — combined with a nonionic surfactant such as polysorbate 20 or 80, which is understood to disaggregate LPS supramolecular structure and strip stabilizing divalent cations. Reconstitution diluents with exactly that composition are common in this market. Where a formulated liquid rather than a lyophilized powder is submitted, we raise LER at scoping and offer a spike-and-hold study, assaying a spiked undiluted sample at intervals out to 7–14 days.

USP derives an endotoxin limit from dose: endotoxin limit = K ÷ M, where K is the threshold pyrogenic dose and M is the maximum dose of product per kilogram of body weight in a single hour. K is 5.0 EU/kg/h for parenteral routes other than intrathecal (equivalently 100 EU/m²/h) and 0.2 EU/kg/h for intrathecal routes. For a 70 kg adult the non-intrathecal case gives a total budget of 350 EU per hour. A material dosed at a maximum of 10 mg/kg/h would give a limit of 5 ÷ 10 = 0.5 EU/mg.

This is why an EU/mg figure alone is uninterpretable. "< 10 EU/mg" on a material used at 1 mg total is 10 EU, trivially inside a 350 EU budget. The same figure on a material used at 500 mg is 5,000 EU, an order of magnitude over. The number is not a safety claim; the number plus the exposure assumption is an exposure calculation, and that calculation belongs to whoever defines the exposure.

For research-use-only material there is no human dose, and therefore no compendially derived limit. Arcadia reports the measured EU/mg and the reporting limit. The client or end-user defines the acceptance criterion appropriate to their research model. Where a report states conformity, it is against a client-supplied criterion, labeled as such, with the decision rule stated. We do not assert or imply a human-use limit.

We also do not use the phrase "pyrogen-free." No test supports it. The defensible construction is: endotoxin not detected at or above X EU/mg by USP ⟨85⟩ kinetic chromogenic method — with X calculated, not assumed.

It is not a sterility test. It detects a molecule, not viable organisms. Material can be non-sterile with undetectable endotoxin, or sterile with high endotoxin.

It does not detect Gram-positive bacteria, fungi, yeasts, or moulds. None of them produce LPS.

It does not detect non-endotoxin pyrogens — Gram-positive cell wall components, lipoteichoic acid, flagellin, and some fungal and viral material are invisible to both LAL and rFC. A passing endotoxin result is not "non-pyrogenic." The Monocyte Activation Test (Ph. Eur. 2.6.30) is the routine method that covers that gap; Ph. Eur. suppressed the rabbit pyrogen test (2.6.8) and revised chapter 5.1.13 with an implementation date of 1 July 2025, directing developers to select an in vitro method on a risk-assessment basis.

It does not detect chemical impurities, degradants, wrong identity, or wrong potency, and it does not detect mycoplasma, viruses, or nucleic acid contamination.

A result is valid only for the dilution tested and only for the container sampled, at the time sampled. It does not describe the lot unless the sampling was statistically designed, and where the client supplied the sample the laboratory has no basis for any statement about the batch.

A number without its controls is not a result. An EU/mg figure with no PPC recovery, no MVD, no λ, and no CSE traceability cannot be evaluated by anyone, and we do not issue one.