Bacterial endotoxin testing, assessed as documentation a purity figure cannot supply

A recall posted this month points at a document this desk has not assessed. On 19 August 2026 Optimal Balance Pharmacy of Houston, Texas announced a voluntary nationwide recall of one lot of Compounded Glutathione 200 mg/mL 30 mL multi-dose vials, to the patient level; FDA published it on 20 August. The stated reason, in the company’s own words on the FDA posting: “pharmacy testing identified elevated bacterial endotoxin levels”.

The material is a tripeptide. The reason given is microbiological, not chemical: the posting records elevated endotoxin and says nothing either way about identity or purity. Endotoxin is a line a chemical certificate cannot carry, because it is not a chemical measurement.

None of this is a template for a research vial. The announcement concerns a compounded prescription preparation dispensed for administration to a person, and the company’s risk statement is written on that basis; the materials covered here are supplied for laboratory research use only and go into nobody at all. As of this writing the page shows no recall classification and no termination notice. What follows assesses the test as a document type, not any supplier.

What the report records

Fields that make an endotoxins test a record rather than a number

Method	Gel clot, photometric or kinetic — the compendial principles
Result	A concentration in endotoxin units, or a pass against a stated threshold
Limit	Calculated per product, by the USP or AAMI methods; FDA withdrew the 1987 guidance and its table
Dilution	The dilution tested, and the maximum valid dilution it sits beneath
Interference	Whether inhibition or enhancement was tested for
Sample basis	One container, or a pooled composite
Lot and date	The production run tested, and when

The reference document is FDA’s Pyrogen and Endotoxins Testing: Questions and Answers, final guidance of March 2026 in a second edition. It is nonbinding on its own face, and points to USP Chapters 85 and 161 and AAMI ST72 for the methods. Its background records FDA accepting a Limulus Amoebocyte Lysate (LAL) test in lieu of the rabbit pyrogen test for more than thirty years.

What it settles

A quantity, against a declared threshold. Failure, on a written rule: per the guidance, an out-of-specification result at the maximum valid dilution that cannot be attributed to testing error means the lot should be rejected; a failure below that dilution is retested higher, and the original stays in the laboratory record. And disputes, in advance. Where a firm substitutes a validated alternative assay for a compendial article, the guidance says the final decision rests on the USP gel-clot method unless the monograph says otherwise. A vial sold for research use is not offered against a monograph, so that tiebreak reaches it only by borrowing.

What it does not settle

Identity and purity. An endotoxins result and a chromatogram do not overlap at any point — the same gap the desk found reading the certificate of analysis as evidence.

A figure without its limit is not a result. FDA says the limit table in its withdrawn 1987 guidance fell out of date as dosage regimes and drug strengths multiplied, and that limits should now come from the USP or AAMI calculations. Its worked example — a veterinary product across species — bases the limit on the maximum dose used in the smallest species; for medical devices the limit follows from what the device contacts. Every anchor the guidance names is a fact about administration, and a research material supplies none of them.

Pyrogens generally. For devices and drug materials the guidance asks firms to assess the risk of non-endotoxin pyrogens, and says the rabbit pyrogen test may be more appropriate where that assessment indicates they may be present.

Its own reliability. Interferences arise from the test article’s physical and chemical properties; where dilution cannot mitigate them the guidance says firms should use the rabbit test. Detection is also affected by storage and handling, so the result belongs to a sample kept a particular way on a particular date. A report silent on inhibition and enhancement has not established that its number means anything.

Which container passed. Finished drug product units may be pooled into a composite — generally small-volume parenterals of 100 mL or less, no more than three units suggested, with the maximum valid dilution divided by the number pooled, because one bad container can be diluted into several good ones.

Reading one, if you are shown one

Seven fields: method, result with units, dilution against the maximum valid dilution, the limit and how it was calculated, one container or a pooled composite, interference assessed or not, lot and date. Endotoxin testing stays a claim until those sit on a retrievable page whose lot number matches the vial. Where no such document exists the question is unmeasured — an untested lot is not a failing lot.

Pros and cons

The bacterial endotoxins test as documentation

Pros

Anchored to compendial chapters a reader can name, not to a phrase on a label
The failure rule is written down, and a retested failure is still recorded
For a compendial article, disputes have a named tiebreak: the gel-clot method decides
Answers a question no chemical assay on a certificate of analysis reaches

Cons

Its limit is calculated from a fact about administration, which a research material does not have
Establishes nothing about identity, purity or content, and cannot be traded against the assays that do
Covers endotoxins only; non-endotoxin pyrogens need a different test entirely
Pooled composites are permitted for finished product units, so a pass can describe several containers, not the one you hold
Needs its own reagents and per-product interference work — a separate assay, not another line on a chromatogram

Verdict

The score sits above the certificate of analysis and below the federal docket. As a specification for material sold by the milligram it arrives incomplete: with no dose or contact to calculate a limit from, the number is comparative rather than a pass mark, and close to meaningless quoted alone.

A certificate of analysis is a chemical document. The recall that opened this piece turned on a microbiological one.