On August 5, 2026, a licensed compounding pharmacy in Austin, Texas voluntarily recalled three lots of an injectable peptide product to the patient level. The recall did not originate with a whistleblower, a counterfeit seizure, or an overseas gray-market vendor. It originated with the pharmacy’s own quality control testing.
Victory Medical Center Pharmacy pulled lots 1980571, 1981940, and 1984345 of its Compounded Glutathione 200 mg/mL multi-dose vials after internal testing identified elevated bacterial endotoxin levels. FDA published the company announcement the same day. The product had already reached consumers in Texas, Florida, and New York, and the pharmacy had already received adverse event reports including fever, chills, severe headache, nausea, vomiting, tachycardia, changes in blood pressure, body aches, and injection site reactions.
FDA’s risk statement was direct: there is a reasonable probability that injectable products with elevated endotoxin levels could cause fevers, hypotension, inflammatory reactions, anaphylactic shock, and death.
Here is what makes this recall analytically important rather than simply unfortunate. Those vials were, in all likelihood, sterile. Sterile filtration is extremely effective at removing bacteria from a solution. It does essentially nothing to remove what bacteria leave behind when they die.
At Vanguard Laboratory, we run identity, purity, potency, and safety panels on peptide material every day. Endotoxin is the test that most surprises the people who submit samples, because it is the one contaminant that survives nearly every procedure a manufacturer would perform to make a product safe. This post explains what endotoxin is, how it is measured, how a genuinely contaminated product can pass adjacent tests, and what a Certificate of Analysis needs to say before an endotoxin claim means anything.
Vanguard does not take a position here on prescribing, clinical use, or the regulatory status of any compound. Our focus is narrower: what the analytical record can and cannot establish.
Glutathione Is a Peptide, and That Matters
Glutathione is frequently discussed as an antioxidant rather than a peptide, which obscures why this recall belongs in a peptide quality conversation. Glutathione is a tripeptide, assembled from glutamate, cysteine, and glycine. It is manufactured, purified, lyophilized or solubilized, filled, and distributed through the same infrastructure that handles every other injectable peptide sold into clinics and med spas.
That shared infrastructure is the point. The failure mode documented in this recall is not specific to glutathione’s chemistry. It is specific to the manufacturing and water-handling history of injectable products in general, and it applies with equal force to BPC-157, TB-500, semaglutide, NAD+, or anything else that arrives in a vial and ends up in a syringe.
What Endotoxin Actually Is
Endotoxins are lipopolysaccharides, abbreviated LPS. They are structural components of the outer membrane of gram-negative bacteria. FDA’s Inspection Technical Guide No. 40 names the usual suspects in pharmaceutical manufacturing: Escherichia coli, Proteus, Pseudomonas, Enterobacter, and Klebsiella.
The molecule has three regions. An O-specific polysaccharide side chain, known as the O-antigen, forms the outermost portion. A core oligosaccharide chain sits beneath it. Anchoring the whole assembly is a lipid region called lipid A, and lipid A is the part responsible for the toxic effects.
Now the detail that reorganizes everything else. FDA’s guidance states that LPS “are not exogenous products of gram negative bacteria” and that “the release of LPS from bacteria takes place after death and lysis of the cell.”
Endotoxin is not a weapon the bacterium fires. It is part of the bacterium’s body. It enters solution when the cell breaks apart.
The consequence is uncomfortable. Every step taken to eliminate microbial contamination — heat, filtration, chemical treatment — kills bacteria, and killing bacteria is precisely the event that liberates endotoxin. A process can drive bioburden to zero while raising the endotoxin load in the same vessel. Thermo Fisher Scientific’s technical guidance on endotoxin testing notes that gram-negative bacteria shed endotoxin throughout their life cycle “but particularly during lytic death where they can release large amounts at once.”
Why Heat and Filters Do Not Solve It
Two properties make endotoxin unusually persistent.
The first is thermal stability. Lipid A is an amphiphilic molecule that tolerates temperatures far outside the range of ordinary sterilization. FDA notes that texts have recommended depyrogenating glassware and equipment by dry heat at 250°C for 45 minutes, and that other reports cite 650°C for one minute or 180°C for four hours as effective. Work published by Tsuji and colleagues in 1978 found that at 170°C, thermal destruction follows second-order kinetics, and achieving a three-log reduction at lower temperatures “might not be practical.”
A standard autoclave cycle runs at 121°C. It sterilizes reliably. It does not depyrogenate.
The second property is that endotoxin is small and soluble enough to pass through the filters used to remove bacteria. FDA states the mechanism without hedging:
Since endotoxins result from high levels of microorganisms, and are not removed by sterilizing or microbiological filters, the subsequent reduction of a high microbiological level will not be associated with a similar reduction of high endotoxin level.
Alternative approaches fare no better. FDA describes filtration, irradiation, and ethylene oxide treatment as having limited effect on pyrogen levels, citing one study in which ethylene oxide exposure achieved roughly an 80 percent reduction in the pyrogenicity of E. coli endotoxin in dialyzers. For an injectable product, an 80 percent reduction from an unacceptable starting point is still an unacceptable product.
There is a further practical hazard. Thermo Fisher notes that lipid A’s hydrophobic anchor readily adsorbs to plastic and, occasionally, glass surfaces of common labware, and that “many products labelled as ‘sterile’ might not be endotoxin-free.” Endotoxin moves through a facility by sticking to things.
How the Limit Is Set
Endotoxin limits are not arbitrary thresholds chosen for convenience. They are calculated from body weight and dose.
FDA expresses the limit as K divided by M. K is 5.0 endotoxin units per kilogram of body weight, which represents the approximate threshold pyrogen dose for humans and rabbits. M is the maximum human dose per kilogram administered in a single one-hour period. For products labeled for intrathecal injection, K drops to 0.2 EU per kilogram, because the central nervous system has far less tolerance.
The worked example FDA provides is instructive in its simplicity. For a non-intrathecal product with a maximum human dose of 10 mL/kg, the limit is 5 EU/kg divided by 10 mL/kg, or 0.5 EU/mL.
| Material or route | Limit basis | Value |
|---|---|---|
| Non-intrathecal drug, general | Threshold pyrogen dose (K) | 5.0 EU/kg/hr |
| Intrathecal drug | Threshold pyrogen dose (K) | 0.2 EU/kg/hr |
| Water for Injection, Sterile Water for Injection, Sterile Water for Irrigation | USP monograph limit | 0.25 EU/mL |
| Bacteriostatic Water for Injection, Sterile Water for Inhalation | USP monograph limit | 0.5 EU/mL |
The water limits deserve attention from anyone reconstituting lyophilized peptide. Bacteriostatic water is held to 0.5 EU/mL under USP. That limit exists because the diluent is part of the injected dose, and its endotoxin contribution adds to whatever the powder brought with it.
How You Measure Something That Is Not Alive
You cannot culture endotoxin. It is a molecule, not an organism, so growth-based methods are structurally incapable of detecting it. The solution the industry arrived at was to borrow a piece of an invertebrate immune system.
In the early 1950s, Frederick Bang observed that the amebocytes — blood cells — of the Atlantic horseshoe crab, Limulus polyphemus, clot in the presence of very small amounts of endotoxin. That observation became the Limulus Amebocyte Lysate assay, and LAL has been the foundation of injectable pyrogen testing ever since. Some formulations use TAL, a comparable lysate derived from Asian Tachypleus species.
FDA recognizes four approaches for end-product release testing: gel-clot, turbidimetric (spectrophotometric), colorimetric, and chromogenic. Gel-clot is qualitative, returning a clot or no clot against a defined sensitivity. Turbidimetric methods track how rapidly the mixture becomes cloudy. Chromogenic methods track color development from a synthetic substrate. The latter two are quantitative, which is what a real Certificate of Analysis requires.
The field is now in transition. In May 2025, the United States Pharmacopeia implemented Chapter <86>, Bacterial Endotoxins Test Using Recombinant Reagents, following early-adoption publication in November 2024. Chapter <86> does not replace or restrict Chapter <85>; it sits alongside it as an additional validated pathway.
Two recombinant reagent classes exist, and the distinction is not cosmetic.
| Reagent | Mechanism | Practical consequence |
|---|---|---|
| Recombinant cascade reagent (rCR) | Reconstructs the full enzymatic cascade found in horseshoe crab blood: Factor C, Factor B, and pro-clotting enzyme | Mimics the natural amplification pathway, supporting broad sensitivity across diverse sample matrices |
| Recombinant Factor C (rFC) | Single-protein mechanism using fluorescence amplification | Does not reproduce the natural enzyme cascade; adoption may require new instrumentation, training, and methodology |
Both eliminate animal-derived material from the supply chain. Neither is a plug-in substitution. Charles River Laboratories describes validation as a process requiring product-specific interference screening, recombinant primary validation, and demonstration of product specificity across matrices, taking months to complete.
Three Ways a Real Failure Disappears on Paper
This is the section that matters most for anyone reading certificates rather than running assays. An endotoxin result can be technically accurate and practically meaningless. FDA’s own inspection guidance documents how.
Pooling conceals single-unit failures. FDA describes a sterile critical device that produced an acceptable endotoxin level from a pooled sample. When extracts of individual units were LAL tested separately, occasional failures appeared. The same pattern emerged with Bacteriostatic Water for Injection: the pooled sample tested below the 0.5 EU/mL limit, individual dosage units exceeded it, and regulatory action followed. Endotoxin contamination is frequently heterogeneous, and averaging is how heterogeneity hides.
Over-dilution manufactures a pass. Every endotoxin method has a maximum valid dilution, calculated from product dose, the endotoxin tolerance limit, and lysate sensitivity. FDA’s language on exceeding it is unambiguous:
Product dilution beyond this determined factor will render a negative result meaningless. Harmful endotoxin concentrations may be diluted below the detectable range of the lysate.
A negative result on an over-diluted sample is not a clean product. It is an unanswered question formatted to look like an answer.
Interference suppresses recovery. Certain products actively inhibit the LAL reaction. FDA identifies chemical inhibitors including chelators such as EDTA, which sequester the divalent cations the cascade requires; protein denaturants; and pH outside the roughly 6.0 to 7.5 window. Physical inhibitors include adsorption of endotoxin onto surfaces and product viscosity. Turbidimetric and chromogenic methods cannot be used with certain turbid or colored products, and precipitate formation can be mistaken for a positive response.
This is why spike recovery validation is non-negotiable. A laboratory demonstrates method suitability by inoculating the actual product matrix with a known low level of endotoxin and confirming it can be recovered. USP <85> expects recovery between 50 and 200 percent; more rigorous laboratories target 75 to 150 percent. Without that validation, a negative result may reflect an inhibited assay rather than a clean sample. FDA further notes that changing lysate reagent source requires revalidation, because reagent manufacturer contributes to variability.
The Regulatory Context Nobody Highlighted
The recall did not arrive in a vacuum. Two weeks earlier, on July 23 and 24, 2026, FDA’s Pharmacy Compounding Advisory Committee met to consider whether seven peptides should be added to the section 503A bulk drug substances list.
FDA’s career reviewers recommended against all seven. Among the specific deficiencies they documented was inadequate impurity and endotoxin testing in available raw material, alongside uncharacterized immunogenicity risk and unresolved questions about salt form. The agency’s reviewer characterized BPC-157, the compound with the most developed evidence base of the group, as not well characterized.
The committee voted to recommend six of the seven anyway. BPC-157, KPV, and TB-500 each passed 8 to 6 with one abstention. MOTS-c passed 7 to 5. Semax and epitalon cleared on the second day. Emideltide was the sole rejection. Those votes are advisory and non-binding, and adding any substance to the list still requires FDA to complete proposed and final rulemaking, a process measured in months to years.
The analytical point stands regardless of where the rulemaking lands. FDA scientists named endotoxin testing of raw material as a gap in July. A licensed pharmacy recalled an injectable peptide product for elevated endotoxin in August. The second event is an illustration of the first.
Market participants have noticed. On an August 10, 2026 earnings call, Hims & Hers CEO Andrew Dudum told investors the company is “currently developing a best-in-class peptides experience, including U.S. manufactured products, clinical-led guidance, and ongoing blood testing,” and that the company had “begun testing for the ingredients in some peptides” to ensure “a safe, verified supply chain.”
Testing is becoming a competitive claim. Which makes it worth knowing what a real endotoxin result looks like.
Endotoxin Is a History Problem
The most useful thing to understand about endotoxin is that it is rarely created at the end of a process. It is usually inherited.
FDA’s inspection guidance is specific on this point: “For parenteral products, inspections have shown that where pyrogen problems were found in dosage forms, and when the source was one of the raw materials, it was the active drug substance.” The pattern was strongest where process water was used at a late stage of synthesis. Endotoxin levels in the drug substance fell when the microbiological quality of that process water improved.
Water systems are the recurring culprit. FDA notes that ambient-temperature Water for Injection systems present the greatest problem, that circulating hot systems above 75°C are least conducive to microbial growth and endotoxin formation, and that reverse osmosis filters “are not absolute” and may need to be installed in series to produce pyrogen-free water. Biologically produced drug substances carry an additional risk: incomplete removal of the production organism during purification leaves endotoxin behind.
So an endotoxin number on a peptide lot is not a snapshot of the vial. It is a record of the water quality, bioburden control, and purification discipline at a facility that may be on another continent and may have shipped the material months earlier.
And it cannot be corrected downstream. FDA’s conclusion is worth quoting exactly:
It is difficult to remove endotoxins from products once present. It is far better to keep finished products and components relatively endotoxin-free rather than have to remove it once present.
What Your Certificate of Analysis Needs to Say
An endotoxin claim is only as good as its specificity. When reviewing a CoA for injectable peptide material, look for five things.
First, a quantitative value with units, expressed in EU/mg for a powder or EU/mL for a solution. The word “pass” without a number tells you nothing about margin. A lot at 90 percent of its limit and a lot at 2 percent of its limit both say “pass.”
Second, the named method and chapter, distinguishing LAL under USP <85> from a recombinant method under USP <86>, and identifying whether the readout was gel-clot, turbidimetric, or chromogenic.
Third, the calculated limit for the product, showing the dose assumption behind it. A result of 3 EU/mg is excellent or unacceptable depending entirely on the intended dose.
Fourth, evidence of method suitability for that specific matrix, meaning spike recovery within an acceptable window and a stated dilution that does not exceed the maximum valid dilution.
Fifth, lot-specific traceability, with the lot on the certificate matching the lot on the vial, and clarity about whether the result came from individual units or a pooled sample.
If those elements are absent, the certificate documents that a test was performed. It does not document that the product is safe to inject.
The Vanguard Standard
Vanguard Laboratory is ISO/IEC 17025:2017 accredited, follows AOAC and USP methodology, and runs instruments 24 hours a day, seven days a week, with an average turnaround under five days for most analyses. We test identity by LC-MS, purity by HPLC-UV, potency by quantitative assay, and safety by ICP-MS for heavy metals and LAL or recombinant methods for bacterial endotoxin. We report results on a clearly stated basis, because a number without a basis is not a result.
The August recall represents the system working, late. A pharmacy tested its own product, found elevated endotoxin, and pulled three lots after patients had already reported reactions. That is genuinely better than the alternative, and the pharmacy deserves credit for acting.
Testing material before it reaches a patient is the same system, working on time.
The question is not whether your supplier is careful. Careful is unverifiable. The question is whether anyone measured, what method they used, and what number they got.
Need endotoxin, identity, purity, or potency testing on a specific lot? Contact Vanguard Laboratory at [email protected] or call 360.967.7010 to discuss a testing panel for your material.
Sources
- U.S. Food and Drug Administration. “Victory Medical Center Pharmacy Issues Voluntary Nationwide Recall of Certain Lots of Compounded Glutathione 200 mg/mL Multi-Dose Vials Due to Elevated Endotoxin Levels.” Company announcement, August 5, 2026.
- U.S. Food and Drug Administration. Inspection Technical Guide No. 40, “Bacterial Endotoxins/Pyrogens.” ORA/ORO/DEIO/IB.
- D’Amico, Dana. “Endotoxins 101: Guide to Bacterial Endotoxin / LAL Testing.” Thermo Fisher Scientific, September 27, 2024.
- Charles River Laboratories. “Guide to USP Chapter <86> for Bacterial Endotoxin Testing.” USP Chapter <86> implemented May 2025.
- Vega, Elena. “Peptides Labeled for Research Only Are Not Approved Medicines and People Are Injecting Them Anyway.” Medical Daily, August 11, 2026.
- Suter, Tara. “Hims & Hers planning to sell compounded peptides.” The Hill, August 11, 2026.
- Tsuji, K. et al. “Dry-Heat Destruction of Lipopolysaccharide.” Applied and Environmental Microbiology, 1978.