Endotoxin and Sterility Testing, Explained

A certificate reporting an HPLC purity figure and a matching observed mass has answered two questions: is this the right molecule, and how much of the sample is it. It has said nothing about whether the material is pyrogenic or sterile. Endotoxin testing is a separate assay, run against a separate compendial chapter, on its own timeline. Two vials can carry identical purity lines and identical mass-match statements and still differ completely in microbial quality.

Why chemical purity and microbial quality are unrelated measurements

Endotoxin is lipopolysaccharide (LPS) from the outer membrane of Gram-negative bacteria. It is a structural fragment, not an organism, so it persists after the organism that shed it is dead. LPS is heat-stable and survives conditions that comfortably kill bacteria, so a process that sterilizes a solution does not depyrogenate it. Sterilizing filtration has the same gap in reverse: a 0.22 µm membrane retains cells, but free LPS passes through it.

So sterile and non-pyrogenic are independent claims, and neither implies the other. A lot can be sterile and still carry a meaningful endotoxin burden from upstream water, raw material, or container contact. It can be endotoxin-clean and still fail sterility from a single late contamination event.

Neither shows up in chemical analysis. HPLC purity is an area-under-curve ratio of UV-absorbing species at the detection wavelength; endotoxin at picogram quantities per milligram of compound does not register as a peak. Mass spectrometry confirms the identity of the intended molecule and is not a survey of everything else in the vial. This is why the microbial section of a certificate of analysis exists as its own block, with its own methods.

USP <85>: the bacterial endotoxins test and its three formats

The Bacterial Endotoxins Test (BET), USP General Chapter <85>, is built on Limulus Amebocyte Lysate (LAL) — a preparation from horseshoe crab hemolymph. In the animal, endotoxin activates a zymogen called Factor C, which primes a serine protease cascade ending in a clot that immobilizes the invader (PMID: 20593268). In vitro, that cascade becomes a detector: trace endotoxin triggers it, and the assay reads out how far it ran. Three compendial formats read it.

Format	What is observed	Readout	Quantitation
Gel-clot	Firm clot at the lysate's labeled sensitivity (λ)	Pass/fail across two-fold dilutions	Semi-quantitative — bracketed between dilutions
Turbidimetric	Onset or rate of turbidity as coagulogen converts to coagulin	Photometric, against a standard curve	Quantitative
Chromogenic	Cleavage of a synthetic substrate releasing p-nitroaniline	Absorbance near 405 nm, against a standard curve	Quantitative

Gel-clot is the referee method: in case of doubt or dispute, the chapter directs that the final decision rest on gel-clot unless the monograph says otherwise. The turbidimetric and chromogenic formats produce a number rather than a bracket, which is why a certificate reporting a specific EU/mg value almost always used one.

The control structure decides whether that number means anything. Every quantitative run must demonstrate positive product control (PPC) recovery: the sample is spiked with a known amount of standard endotoxin, and the assay must recover it within the compendial window. If it does not, the matrix is inhibiting or enhancing the cascade and the reported value is unusable. Peptide and protein matrices interfere often enough that this is routine. The remedy is dilution, bounded by the maximum valid dilution (MVD) — past which the assay can no longer see the relevant limit.

Recombinant Factor C: what it is and why the alternative exists

Recombinant Factor C (rFC) replaces crab-derived lysate with a single recombinantly expressed protein — the cascade's primer — paired with a fluorogenic substrate. Because only Factor C is present, the cascade branch that responds to (1→3)-β-D-glucans is absent, removing a known source of false positives. It also removes an animal-derived raw material from the supply chain, which is why the reagent was developed (PMID: 20593268).

The comparison literature is mixed, and it matters which way each study cut and who ran it. Bolden and Smith, at Eli Lilly, validated an endpoint-fluorescence rFC method across several pharmaceutical products and reported it equivalent or superior to the compendial BET (PMID: 28733334). A Sanofi Pasteur group compared two LAL and two rFC assays across four complex vaccine matrices and found rFC at least equivalent — though a glucan-containing product suited LAL less well, and a protease-containing one suited rFC less well once those proteases were inactivated (PMID: 32179709).

The counterweight comes from Charles River Laboratories, which manufactures LAL reagents — an affiliation worth weighing. They tested 128 samples of naturally occurring environmental endotoxin drawn from pharmaceutical water systems rather than purified reference standard. Non-inferiority held between the two compendial LAL methods but could not be claimed for any recombinant reagent evaluated (PMID: 33454378). Environmental LPS is structurally heterogeneous in a way a purified standard is not, and the reagents do not respond identically.

So rFC on a certificate is not a defect, but it is a different method with a different specificity profile, and it should be named rather than reported generically as "LAL." "Recombinant" alone is underspecified too — most rFC reagents carry only the first enzyme of the crab cascade, while at least one includes all three. The European Pharmacopoeia has carried a dedicated rFC chapter for several years, and USP addressed recombinant reagents in a separate chapter rather than folding them into <85>.

USP <71> sterility testing runs on a different clock

Sterility is a growth test, not an instrument reading. Under USP <71>, sample units are either filtered through a membrane that is then transferred to media, or inoculated directly into media, then incubated for 14 days in two systems: fluid thioglycollate medium at roughly 30–35 °C for anaerobes and aerobes, and soybean–casein digest medium at roughly 20–25 °C for fungi and aerobes. Nothing shortcuts that incubation. A lot released before day 14 was not released on a completed sterility test.

Its failure modes:

• It is a sampling test, not a batch guarantee. A defined number of units is tested, and a pass is a statistical statement — reliable against high contamination rates, weak against sparse low-level contamination. That is inherent to the method, not a lab defect.
• The article itself can suppress growth. Bacteriostatic or fungistatic character can inhibit the organisms the test is looking for, which is why method suitability must be demonstrated first.
• False positives happen. Contamination introduced during testing produces growth unrelated to the lot, and resolving it takes a documented investigation, not a quiet retest.
• Bioburden is not sterility. A line reading "total aerobic microbial count <10 CFU/g" is an enumeration under USP <61>, with <62> covering tests for specified organisms. It counts what grew — useful, but not a sterility test, and the two are frequently conflated on supplier certificates.

How to read an EU/mg endotoxin line item

The compendial endotoxin limit for a parenteral article is not fixed. It is calculated as K/M, where K is the threshold pyrogenic dose per kilogram per hour for the route in question and M is the maximum amount of that article delivered per kilogram in the same hour. K is a published constant; M is entirely a property of the protocol — so a supplier of a research compound cannot compute an applicable limit on anyone's behalf.

That is why certificates report a per-mass specification instead, commonly <1 EU/mg. This is a conservative supplier-side ceiling, not a compendial limit, and it scales with the mass in the vial. The distinction matters most on high-mass presentations: a 5 mg vial and a 1500 mg glutathione vial held to the same <1 EU/mg specification carry total endotoxin ceilings three hundred-fold apart. A diluent is a separate article with its own release testing, which is why bacteriostatic water is certified on its own rather than inheriting the compound's certificate.

Four things make an endotoxin line informative:

1. The method is named — gel-clot, turbidimetric, chromogenic, or recombinant, not just "LAL."
2. The lysate sensitivity (λ) and the dilution are stated. A "<1 EU/mg" result obtained near the MVD is a weaker claim than the same number obtained close to neat.
3. Control recovery is reported. Without a passing positive product control, a low reading and an inhibited assay look identical on paper.
4. The result belongs to this lot. A batch-typical figure carried forward across lots is not a measurement of the vial in front of you.

Merit lots are compounded in an ISO-certified US facility to USP <797>, and each certificate carries the microbial section alongside the chromatogram and spectrum, not in place of them. Matching lot numbers, confirming the issuing laboratory, and spotting sections marked "not tested" are covered in how to verify a certificate of analysis.

Read the endotoxin line for its method, its dilution, and its controls — not just its number — and read the sterility line as a separate 14-day growth test that a bioburden count does not substitute for.

For research use only. Not for human or veterinary use. This article is interpretive guidance; the certificate of analysis is the authoritative release document for any given lot.