Endotoxin Testing for Peptide Buyers: Read Past 'Pass'

A certificate line reading '<0.5 EU/mg — Pass' reports a dilution calculation, not a verified property of the lot. Four additional fields per lot make the result auditable, and a contractual right to demand a hold-time study covers what those four fields still cannot settle.

A peptide certificate of analysis that reports endotoxin at all usually reports it in one line: a chapter reference, a numeric limit, and the word Pass. "<0.5 EU/mg, USP <85>, Pass" is the canonical form, and purchasing departments treat it as settled. It is not settled. It is the output of a dilution calculation whose inputs are not printed anywhere on the page.^2,3

FDA's draft guidance on setting endotoxin limits during product development frames an endotoxin specification as three distinct elements — the test, the method, and the acceptance criteria — rather than a single pass/fail figure, and recommends product-specific suitability testing. A one-line pass discloses the acceptance criterion and nothing else. The other two elements, plus the bench conditions that produced the number, are what belongs on every incoming lot record and what this article argues a buyer should require before a lot is accepted.³

The Pass/Fail Line That Isn't One

An endotoxin result is never measured in the vial as supplied. It is measured in a diluted aliquot and multiplied back out. The smallest concentration the assay can resolve in the original material is therefore the reagent's labeled sensitivity multiplied by whatever dilution the lab chose. Dilute further and the detection floor rises in lockstep. A reported "less than" figure can be the floor the lab created rather than a ceiling the lot earned.²

That makes "<0.5 EU/mg — Pass" ambiguous between two very different lots. One was examined across the full sensitivity of the assay and showed nothing. The other was diluted until the assay could no longer see below the limit, and then showed nothing. Both certificates read identically. Nothing in the standard pass/fail line distinguishes them, and no amount of re-reading the certificate will.^2,3

This is why FDA's three-element framing matters commercially rather than academically. The acceptance criterion alone is unauditable. Without the method and the conditions under which it was run, a receiving buyer has no way to compare one lot against the next, no way to compare two suppliers quoting the same catalog item, and no record that survives a downstream question about why a particular lot was released.³

What USP <85> Actually Permits at the Bench

The maximum valid dilution is the compendial ceiling: the greatest dilution at which the endotoxin limit for that material can still be validly determined. It is computed from the endotoxin limit, the concentration of the sample as tested, and lambda — the labeled sensitivity of the lysate lot in use. Change the reagent lot and the MVD changes. Change the sample concentration and the MVD changes again. It is a per-lot, per-reagent number, not a catalog constant.²

FDA's Guidance for Industry on Pyrogen and Endotoxins Testing — the agency's questions-and-answers document — describes what the chapter permits when a test does not come out clean. A failed test may be repeated at greater dilution, up to the maximum valid dilution. If interference persists past that point, samples may be tested individually rather than pooled. Both moves are legitimate. Both are invisible on a certificate that prints only the final line.²

• A first run showing interference, repeated at a higher dilution that resolves it, and reported as a single passing result
• A pooled sample that would not resolve, broken into individual samples, with only the aggregate conclusion reaching the certificate
• A run conducted at or near the MVD, where the detection floor sits at the limit itself and anything below the limit is structurally undetectable
• A run conducted well inside the MVD, with real sensitivity margin — reported in exactly the same words as the run above²

Each of those is a different quality signal, and all four reduce to "Pass." The buyer is not being deceived. The buyer is being handed the conclusion of a judgment call without the judgment call. For a wholesale account placing repeat orders against the same SKU, that means lot-to-lot consistency is being assumed from a field that cannot demonstrate it.²

An Enforcement File That Shows the Gap in Practice

This is not a theoretical weakness in certificate design. FDA has already written it up. The FDA warning letter archived at fda.gov/media/86611 cites a finished-drug manufacturer for failing to calculate the maximum valid dilution for finished product samples tested under USP <85> — the compendial limit at which the endotoxin result can be validly determined. The deficiency was not a failing endotoxin number. It was the absence of the calculation that tells anyone whether the number meant anything.¹

The same letter cites the manufacturer for running a bacterial endotoxins test with no positive product control sample included in the run. A test with no positive product control has not demonstrated that the sample matrix permitted detection at all. The result is a reading, not a finding.¹

Both cited deficiencies would have been completely invisible to a customer holding the resulting certificate. The certificate would have said Pass. The gap the inspector found — no MVD calculation, no PPC — lives entirely in the laboratory record the buyer never receives. That is the practical case for demanding those fields on the document instead of trusting that they exist somewhere upstream.^1,3

The Positive Product Control and the 50–200% Window

The positive product control is the formal test for whether the material itself is interfering with detection. A known quantity of endotoxin standard is spiked into the sample as tested, and the assay's recovery of that spike is calculated as a percentage. Under compendial bacterial endotoxin testing, a sample is considered free of interfering factors when that recovery falls between 50% and 200%.⁴

That window is wide on purpose, and its width is the point for a purchasing file. A recovery of 55% and a recovery of 190% both clear. Both are reported, on a certificate that compresses them, as the same thing. A lot trending toward the edge of the window across successive runs is a matrix behaving differently over time, and a buyer tracking recovery as a number will see that drift three lots before a buyer tracking "PPC: Pass" sees anything at all.⁴

Peptide formulations are the exact context where this field earns its place. Bulking agents, surfactants, and buffer salts are standard formulation components and are standard interference sources. A certificate that confirms the spike recovered inside the window has told the buyer that detection was possible. A certificate that omits it has told the buyer nothing about whether detection was possible, which makes the headline result unfalsifiable.⁴

Masked Endotoxin: Why a Passing PPC Isn't the End of the Question

Low endotoxin recovery (LER) is documented in the peer-reviewed literature on masked endotoxin as a phenomenon in which chelators and detergents common in protein and peptide formulation buffers mask lipopolysaccharide from detection by LAL-based assays. The lipopolysaccharide is present. The assay does not see it. The certificate reports what the assay saw.¹⁰

The consequential finding for anyone building an incoming-goods specification is this: LER-driven masking can occur even while the positive product control shows no evidence of test interference. A clean recovery inside the 50–200% window does not, on its own, rule out masking. The spike and the native material do not necessarily behave the same way in the same buffer over the same interval, and the control is calibrated on the spike.¹⁰

The peer-reviewed literature on masked endotoxin documents a specific response to this problem, and it is not a reread of the certificate. Where low endotoxin recovery is suspected, the documented response is a hold-time study of endotoxin in addition to standard pharmacopoeial testing requirements — spiked material held in the actual formulation matrix over time, with recovery measured across the interval rather than at a single point.¹⁰

Recombinant Factor C Under USP <86>: Alternative, Not Default

USP <86>, the Bacterial Endotoxins Test Using Recombinant Reagents, is an FDA-recognized consensus standard that uses recombinant Factor C or recombinant cascade reagents to detect or quantify endotoxin. Certificates citing rFC are increasingly common as reagent supply shifts, and a receiving buyer will encounter them whether or not a policy exists for handling them.⁶

The classification matters more than the chemistry. USP <86> recombinant-reagent methods are classified as alternative tests, not default replacements for USP <85>, unless a specific monograph states otherwise. A supplier switching from <85> to <86> mid-relationship has changed the method element of the specification, and under FDA's three-element framing that is a specification change, not a footnote.^6,3

FDA's Pyrogen and Endotoxins Testing Q&A guidance is explicit about what a switch obligates. Where a manufacturer uses a recombinant Factor C-based endotoxin assay, method validation must follow the Bacterial Endotoxins Test chapter's photometric quantitative technique requirements together with the compendial procedure validation chapter. That validation package exists or it does not, and the question of whether it exists is answerable before a shipment is booked rather than after it lands.²

1. Decide, in writing, whether rFC results under USP <86> are acceptable on incoming lots at all, or acceptable only alongside a <85> result
2. If accepted, require the method validation package referenced in FDA guidance to be named and dated on the certificate or in the supplier's document index
3. Require advance written notice of any method change between <85> and <86> for an established SKU, treated as a specification change rather than a reagent substitution
4. Record which method produced each historical lot result, so trend comparisons are not drawn across two different methods^6,2

Rewriting the Incoming-Goods Spec: Four Fields Per Lot

Everything above converts into four fields that a quote request can demand and a receiving clerk can check without a chemistry background. None of them require the supplier to generate new data. All four already exist in the laboratory record of any run that was performed correctly, which is precisely why a supplier's inability to produce them is informative.^3,2

Two drafting notes make the clause enforceable. First, require the four fields per lot, not per product — lambda and MVD move with the reagent lot, so a product-level attestation issued once a year satisfies nothing. Second, require them as values, not as verdicts: "PPC recovery 112%" is a field, "PPC acceptable" is a restatement of the pass line this specification exists to replace.^4,3

A supplier that can return all four within a normal quote cycle is reading from a laboratory report. A supplier that returns them only after a delay long enough to assemble them is sourcing from a contract lab it does not have a document channel into. That distinction is worth knowing before a purchase order is signed, and it costs one email to establish.³

When to Hold Payment and When to Demand a Hold-Time Study

A missing field and a suspected masking problem are different events requiring different responses. A missing field is a documentation gap: the lot goes to quarantine, acceptance is withheld, and the supplier is asked to produce the laboratory record. Most of these close within days, because the data existed all along and only the transcription to the certificate was lazy.³

Suspected low endotoxin recovery is the escalation. It cannot be closed by producing the same certificate with more numbers on it, because the four fields can all be present and correct while masking proceeds undetected. The documented response in the masked-endotoxin literature is a hold-time study — recovery measured across an interval in the actual formulation matrix — requested in addition to standard pharmacopoeial testing.¹⁰

• The formulation includes a chelator or a detergent, the two component classes documented as driving low endotoxin recovery in protein and peptide buffers
• PPC recovery is reported inside the window but trending toward an edge across consecutive lots of the same SKU
• Endotoxin results read as non-detect across every lot ever supplied, with no variation at all, in a matrix where variation would be expected
• The method changed between <85> and <86> without a corresponding update to suitability data
• A run was performed at or very near the calculated MVD, leaving no sensitivity margin below the limit^10,2

The contractual point is that a hold-time study takes weeks and cannot be commissioned usefully after a lot has been accepted, paid for, and partially distributed. No standard incoming-goods specification in this trade reserves the right to demand one. Adding a clause that permits the buyer to require a hold-time study on a named lot, before acceptance, with the acceptance clock paused while it runs, is the only provision that reaches the failure mode the four fields leave open.^10,3

Priced against the alternative, the clause is cheap. The cost of invoking it is a delayed shipment on one lot. The cost of not having it is a distributed lot whose endotoxin record consists of a line that reported a dilution calculation, and a supplier under no obligation to go back and check.³