Endotoxin Testing: What The LAL Assay Measures, And What It Does Not

Endotoxin is not an organism. It is a structural fragment: lipopolysaccharide from the outer membrane of Gram-negative bacteria, which survives conditions that kill the bacterium that made it. Autoclaving at 121 °C for 15 minutes reliably kills cells and leaves lipopolysaccharide largely intact, which is why endotoxin is handled as a separate question from microbial presence and needs its own assay and its own removal step.

That assay, in its standard form, is the Limulus amebocyte lysate test described in USP General Chapter <85> and in European Pharmacopoeia 2.6.14. Lysate prepared from the blood cells of the horseshoe crab contains a protease cascade that lipopolysaccharide activates. Depending on the method chosen, the readout is a visible gel, a rise in turbidity, or the release of a coloured fragment from a synthetic substrate.

The unit is the endotoxin unit, defined against a reference standard endotoxin rather than by mass, because different lipopolysaccharide structures are not equally potent. One endotoxin unit corresponds to roughly 100 picograms of the current reference standard preparation. A figure in nanograms per millilitre, with no standard named, is not a comparable number.

The three compendial methods, and what each returns

Gel-clot is a limit test: it answers whether the sample exceeds a stated concentration at a stated dilution, and nothing finer. The kinetic methods are quantitative against a standard curve, usually four points spanning two to three orders of magnitude, with a correlation coefficient of at least 0.980 required for the curve to be accepted. Recombinant factor C uses a cloned single enzyme instead of the full lysate cascade and is recognised in European Pharmacopoeia 2.6.32.

Why the sample always has to be diluted

Almost nothing is tested neat. Peptide solutions, buffers and excipients interfere with the cascade in both directions, and the way that interference is managed is by dilution to a point where the sample no longer distorts a known spike. That point is bounded: every lysate has a maximum valid dilution, calculated from the endotoxin limit for the material and the labelled sensitivity, beyond which the assay can no longer see the concentration of interest.

Validity therefore rests on two controls run alongside the sample. A negative water control confirms the reagents and labware are clean, using water meeting the requirements of USP <1231> for bacterial endotoxins. A positive product control, the sample spiked with a known endotoxin addition, must recover between 50% and 200% of the spike. Recovery outside that window invalidates the result for that sample at that dilution.

The failure mode: low endotoxin recovery

The characteristic failure is low endotoxin recovery, usually shortened to LER. A sample is spiked with a known quantity of endotoxin, held, and then tested, and the spike has become partly or wholly undetectable. Recoveries falling from near 100% to below 20% over a 24 hour hold have been reported repeatedly in formulations combining a chelator such as citrate or phosphate with a non-ionic surfactant such as polysorbate 20 or polysorbate 80.

The accepted explanation is physical rather than chemical: lipopolysaccharide aggregates are broken up and the resulting monomers are shielded by surfactant micelles, so the cascade no longer sees them. The endotoxin has not been destroyed. It is simply invisible to the assay in that matrix. LER shows up only if a hold-time spike study is actually performed, which is why a bare endotoxin figure with no hold-time data behind it carries less information than it appears to.

The mirror-image problem is interference from beta-glucans. Cellulosic filters and some cellulose-derived excipients release glucans that activate the factor G branch of the lysate cascade, producing a response with no endotoxin present at all. Glucan-blocking buffers, or recombinant factor C reagents that lack the factor G branch, are the usual answers.

What the number cannot tell anyone

• Sterility. Endotoxin can be low in material carrying viable organisms, and high in material with nothing alive in it. The two questions are measured by different chapters and different methods.
• Identity. The cascade responds to lipopolysaccharide. It is indifferent to whether the peptide in the vial is the sequence on the label.
• Purity. Synthesis-derived impurities such as truncated sequences and oxidised residues are invisible to the assay.
• Container history. A result generated on one aliquot says nothing about what entered the vial after the aliquot was drawn.

Those boundaries are the reason endotoxin, bioburden and sterility are reported separately in a pharmaceutical specification rather than collapsed into one line. Each covers a different failure route, and passing one carries no implication for the others.

Where we stand on this

Sterility, endotoxin and pyrogen testing are not performed and not claimed for any material Aurum lists. What Aurum does publish is identity and purity data: purity independently assayed by reverse-phase HPLC, and identity assayed by mass spectrometry. Those are different measurements answering a different question, and neither one substitutes for a bacterial endotoxin test. An endotoxin figure is not among the specifications Aurum publishes.

Stating that plainly is more useful than an unsourced number would be. An endotoxin figure means something only when the method, the lysate sensitivity, the dilution, the positive product control recovery and the hold time are all stated with it. Everything short of that is an adjective.