SOP: Bacterial Endotoxin Testing (BET) by Gel-Clot and Photometric Methods

This is a ready-to-use SOP for routine bacterial endotoxin testing. Replace every <<FILL: ...>> placeholder with your own specifics, set your document numbers and dates, and route it through your normal document control, review, and approval. A worked filled specimen follows the template. The compendial chapters referenced here are copyrighted; this procedure describes them in its own words and states only the numeric criteria. Read the current chapter text before you rely on any of it. This is educational reference content, not regulatory advice.

Document control header

1. Purpose

This procedure defines how <<FILL: COMPANY NAME>> performs, controls, and reports the bacterial endotoxin test (BET) on <<FILL: products, in-process samples, water, and components in scope>>. The objective is that every reported endotoxin result was produced by a run whose controls demonstrate the assay was working and the sample matrix did not hide or exaggerate endotoxin.

2. Scope

This procedure covers gel-clot (limit test and endpoint titre) and photometric (kinetic or endpoint turbidimetric and chromogenic) endotoxin testing performed at the laboratory named in the header. It applies to release testing, in-process testing, water testing, and component rinse testing.

It does not cover:

• Derivation of the product endotoxin limit and maximum valid dilution, which is performed once per product on the endotoxin limit and MVD calculation worksheet and referenced here as a fixed input.
• Product-specific method suitability (interference) validation, governed by <<FILL: SOP-ID for compendial method verification>>. See also compendial method verification.
• Low endotoxin recovery hold-time studies, governed by <<FILL: protocol or SOP reference>>. See the LER hold-time study protocol.
• Depyrogenation cycle qualification, governed by <<FILL: SOP-ID>>. See depyrogenation and dry heat sterilization.

2.1 Method family in use

State which compendial chapter the laboratory operates under for each product. USP <85>, Bacterial Endotoxins Test, covers the amebocyte lysate methods (gel-clot, turbidimetric, chromogenic). USP <86>, Bacterial Endotoxins Test Using Recombinant Reagents, is a separate chapter covering recombinant cascade and recombinant Factor C methods; it became official on 1 May 2025. The two chapters sit alongside each other. Selecting one is a documented decision recorded per product in <<FILL: method register>>, not a default, and where the method is registered in a marketing application a change is a variation in that market.

3. Responsibilities

4. Definitions

• EU: Endotoxin Unit, the activity unit, defined against the reference standard endotoxin (RSE) and carried into routine use through a control standard endotoxin (CSE) with documented traceability.
• lambda: the labelled sensitivity of the lysate in EU/mL. For gel-clot it is the lowest concentration that reliably produces a clot; for photometric methods it is the lowest point on the standard curve.
• LAL reagent water (LRW): water shown to be free of detectable endotoxin at the sensitivity of the assay and free of interfering substances. Treated in this procedure as a critical reagent.
• MVD: maximum valid dilution, the greatest dilution at which the assay can still detect endotoxin at the product endotoxin limit.
• PPC: positive product control, the sample spiked with a known endotoxin amount and carried through the same preparation as the sample, used to demonstrate the matrix does not inhibit or enhance the reaction.
• Invalid run: a run in which one or more control criteria in section 6 failed, so no sample result from that run may be reported.

5. Procedure

5.1 Reagent and labware control

1. Use only depyrogenated glassware, or single-use labware certified endotoxin-free by the supplier at or below <<FILL: e.g. 0.005 EU/mL or 0.05 EU per item>>. Record the depyrogenation cycle reference or the supplier certificate lot for every run.
2. Glassware depyrogenated in house is processed on a qualified dry heat cycle demonstrating not less than a 3.0 log10 reduction of endotoxin. Confirm the cycle qualification is current before use; a lapsed qualification makes the labware unqualified.
3. Treat LAL reagent water as a critical reagent. Record the lot on every run. Do not decant into non-depyrogenated vessels, do not return unused water to the bulk container, and discard opened containers after <<FILL: e.g. 14 days>> or the supplier limit, whichever is shorter.
4. Reconstitute lysate with the volume and diluent stated by the supplier, swirl rather than vortex unless the supplier directs otherwise, and use or store reconstituted lysate strictly within the labelled window and temperature.
5. Do not mix lysate lots, standard lots, or water lots within a single run.
6. Record every reagent lot, expiry, and receipt or reconstitution date on the run record (section 8).

5.2 Lysate lot confirmation of labelled sensitivity

Before a new lysate lot is used for reportable testing, confirm its labelled sensitivity.

1. Prepare a standard series in LRW at 2 lambda, lambda, 0.5 lambda, and 0.25 lambda from the qualified CSE.
2. Test each concentration in <<FILL: number, e.g. 4>> replicates, plus a negative control.
3. Determine the endpoint for each replicate series and calculate the geometric mean endpoint.
4. Acceptance: the geometric mean endpoint falls between 0.5 lambda and 2 lambda. Outside that range, the lot is not confirmed at its labelled sensitivity and is not released for use; quarantine it and notify the supplier and the supervisor.
5. Record the confirmation on <<FILL: lysate qualification form ID>> and file it against the lot. Repeat this confirmation for every new lysate lot, and whenever the CSE lot changes.

For photometric methods, the equivalent lot control is the standard curve in section 5.3, run with each assay rather than once per lot.

5.3 Standard curve requirement, photometric methods

1. Prepare a standard series in LRW spanning the range required for the sample dilutions in the run, using not fewer than three concentrations.
2. Run the standards in replicate on the same plate as the samples. A curve carried over from a previous run is not acceptable.
3. Acceptance: the absolute value of the correlation coefficient of the log-log regression is not less than 0.980. Below that, the run is invalid.
4. The lowest standard defines the assay sensitivity for the run; a sample result reported below it is reported as less than that value, never as a number.
5. Confirm the curve range covers the expected sample values after dilution correction. A sample reading above the highest standard is diluted further, within MVD, and re-run; it is not extrapolated.

5.4 Gel-clot: the four control types

Every gel-clot run contains four tube types, and each answers a different question.

A negative sample tube alongside a negative PPC is not a pass. It is an invalid test, because the run has just demonstrated the matrix can suppress a known spike.

5.5 Gel-clot execution

1. Confirm the product endotoxin limit, lambda, and MVD from the approved worksheet for that product. Do not recompute them at the bench.
2. Prepare the sample at the routine dilution stated in the test method, which is at or below MVD.
3. Where pH adjustment is part of the validated preparation, adjust into the range <<FILL: e.g. 6.0 to 8.0>> using endotoxin-free reagents and record the measured pH.
4. Dispense equal volumes of lysate and test solution into each tube per the supplier instructions, mix gently, and avoid introducing bubbles.
5. Incubate undisturbed at 37 +/- 1 C for 60 +/- 2 minutes. Do not open, move, or vibrate the block during incubation.
6. At the end of incubation, invert each tube through a full 180 degrees in one steady movement. A firm gel that stays in place is positive. No gel, or a gel that collapses, is negative.
7. Read every tube once, at the end of the incubation window. Re-reading a tube later, or reading before the window closes, is not permitted and the result may not be recorded.
8. Record the result for every tube immediately, including the controls, on the run record.

5.6 Photometric execution

1. Confirm the product endotoxin limit, lambda, and MVD as in section 5.5 step 1.
2. Prepare the standard series, negative control, sample dilutions in replicate, and a PPC for each distinct sample matrix in the run.
3. Load the plate per the approved plate map. Record any deviation from the map before starting the read.
4. Run the assay on the qualified reader using the approved protocol file <<FILL: protocol file name and version>>.
5. Verify the run controls in section 6 before looking at sample values. This order is deliberate: it prevents a control criterion being reinterpreted after a sample result is known.
6. Record the instrument file identifier and archive the raw data file per <<FILL: data retention SOP>>.

5.7 Run validity criteria

A run is valid only when all of the following are true. Any failure invalidates the whole run, not the individual well or tube.

5.8 Interference resolution within MVD

Where a PPC fails, the matrix is interfering and the result cannot be reported. Resolve interference in this order and re-validate the method suitability afterwards.

1. Dilute further, up to but not beyond MVD. This is the first and cleanest option because it needs no chemistry and the headroom is already defined.
2. Adjust pH into the range validated for the method, using endotoxin-free acid or base.
3. Use a divalent cation buffer or a formulation-specific diluent where the interference is a cation or chelation effect.
4. Change to an endotoxin-specific reagent formulation where beta-glucan reactivity is the suspected cause.
5. Where interference cannot be resolved at or below MVD by any of the above, the product cannot be tested by this method. Escalate to the method owner; the options are an alternative method or, where non-endotoxin pyrogens cannot be excluded, a different pyrogen test entirely. Record the decision.

Do not resolve interference by testing beyond MVD. A dilution beyond MVD cannot detect endotoxin at the product limit, so a negative result at that dilution carries no information about compliance with the limit.

5.9 Result calculation and dilution correction

Photometric. The instrument reports the concentration in the tested dilution. Multiply by the dilution factor to obtain the concentration in the undiluted product.

Reported result (EU/mL) = instrument value (EU/mL) x dilution factor

Compare the corrected result against the product endotoxin limit. Round only at the final step, to <<FILL: number>> significant figures, per <<FILL: rounding SOP>>.

Gel-clot limit test. The result is an attribute, not a number. A negative sample tube at a dilution at or below MVD means the endotoxin concentration is below the product limit at that dilution. Report as “less than” the concentration the dilution corresponds to, and record the dilution used.

Gel-clot endpoint titre. Where a value is needed, run a dilution series, take the geometric mean endpoint across replicates, and multiply by the dilution factor at that endpoint.

Record the arithmetic on the run record so a reviewer can reproduce it without recalculating from memory. Where the result feeds a batch disposition, a second person verifies the dilution correction and signs.

5.10 Invalid-run handling

1. When any criterion in section 5.7 fails, mark the run invalid on the run record before any sample result is transcribed. Do not report, trend, or discuss sample values from an invalid run as if they were results.
2. Record the failed criterion, the observed value, and the samples affected.
3. Notify the supervisor the same working day.
4. Retain the raw data for the invalid run. Invalid runs are archived, not deleted, and the audit trail of the acquisition file is retained.
5. Investigate the assignable cause before repeating: reagent lot, water lot, labware, pH, incubation temperature, instrument, technique. Record the cause or record that no cause was found.
6. Repeat the test on a fresh sample aliquot only after the cause is addressed, or under a documented investigative plan approved by the supervisor.
7. An invalid run is not an OOS result, but repeatedly invalidating runs on the same sample without a recorded cause is itself a signal. After <<FILL: number, e.g. 2>> consecutive invalid runs on one sample, escalate to QA before testing again.

5.11 OOS routing

1. A valid run whose dilution-corrected result meets or exceeds the product endotoxin limit is an out-of-specification result. Report it as such the same working day.
2. Do not repeat the test, dilute further, or test a second sample before the OOS process is opened. Retesting into compliance without an approved investigation is the single most cited failure in this area.
3. Route the OOS per <<FILL: SOP-ID for OOS investigation>> and follow the phase 1 laboratory investigation and phase 2 logic described in the OOS investigation process.
4. Quarantine the affected material and notify Manufacturing and QA. The result stands until the investigation concludes otherwise on documented, scientific grounds.
5. Where the affected sample is an in-process bioburden or endotoxin control point, assess upstream impact including water systems and hold times.
6. Do not average a failing result with a passing one, and do not exclude a replicate without a documented assignable cause.

6. Acceptance criteria

A reportable endotoxin result is acceptable when all of the following are true.

• All run validity criteria in section 5.7 were met, and the evidence is on the run record.
• For photometric methods, the standard curve absolute correlation coefficient was not less than 0.980 and the PPC recovery fell between 50 and 200 percent.
• For gel-clot, the lysate lot geometric mean endpoint was confirmed between 0.5 lambda and 2 lambda, incubation was 37 +/- 1 C for 60 +/- 2 minutes, the negative control was negative, and both the positive control and the PPC were positive.
• The tested dilution was at or below the approved MVD for that product.
• The dilution-corrected result was calculated using the approved limit and was verified by a second person where it feeds a disposition.
• Any invalid run was recorded, investigated, and retained, and any OOS result was routed before repeat testing.

7. References

21 CFR 211.165 (testing and release for distribution) and 211.167 (special testing requirements for products purporting to be sterile and pyrogen-free). 21 CFR 610.9 and 610.13(b) for biological products, where a pyrogen test requirement and its waiver route apply. USP <85> Bacterial Endotoxins Test, and USP <86> Bacterial Endotoxins Test Using Recombinant Reagents (official 1 May 2025). USP <1085> Guidelines on the Endotoxins Test; USP <1225> Validation of Compendial Procedures. Ph. Eur. 2.6.14 Bacterial endotoxins, which now carries the recombinant Factor C fluorimetric end-point method as Method G; Ph. Eur. 2.6.32 Test for bacterial endotoxins using recombinant factor C, the standalone chapter being suppressed as its content transfers into 2.6.14, with implementation from 1 January 2027, so cite 2.6.14 Method G in documents intended to remain in force past that date; Ph. Eur. 5.1.13 Pyrogenicity (in force 1 July 2025). FDA, Pyrogen and Endotoxins Testing: Questions and Answers (Edition 2), March 2026. EU GMP Annex 1, Manufacture of Sterile Medicinal Products.

Confirm the current version and clause numbers of each reference before issue. The compendial chapters are copyrighted; consult the chapters themselves for the method text.

8. Record generated: BET run record

9. Revision history

10. Approvals

Filled specimen

The following shows the run record completed for an example monoclonal antibody drug product tested by kinetic chromogenic assay, and a second gel-clot example on the same page for contrast. Company, product, and numbers are illustrative; replace them with your own.

Example 1: photometric, valid and passing.

Example 2: gel-clot, invalid run and what happened next.

Run BET-2026-0421, same product family, gel-clot with lambda 0.125 EU/mL, tested at 1:40. Negative control negative, positive control positive, sample negative, PPC negative. The analyst marked the run invalid before transcribing the sample result, recorded the failed criterion as PPC negative at 2 lambda, and notified the supervisor the same day.

Investigation found the sample had been prepared at pH 4.9, outside the validated 6.0 to 8.0 range, because a buffer had been omitted from the preparation. The cause was assignable and documented. The sample was re-prepared with the buffer at pH 7.1 and re-tested at the same 1:40 dilution; the PPC was positive and the sample negative, and the result was reported.

The sequence matters more than the outcome. The PPC failure was recognised as an invalid run rather than a passing sample, the run was marked invalid before the sample value was written down, the raw data was retained, and the cause was found before the repeat. Had no cause been found, the repeat would have needed supervisor approval under a documented plan, and two consecutive invalid runs would have gone to QA.

Example 3: OOS routing. Run BET-2026-0433, water for injection point of use SP-12, kinetic chromogenic, all controls valid, dilution-corrected result 0.31 EU/mL against the 0.25 EU/mL limit for Water for Injection. The analyst reported the OOS the same working day, did not retest, and the OOS was opened under the laboratory investigation procedure. The point of use was locked out and the sanitisation and sampling history reviewed as part of the phase 2 assessment.

Common inspection findings this SOP prevents

• A sample reported as passing on a run where the positive product control failed, so the matrix was known to suppress a spike and the negative sample result was reported anyway.
• Testing at a dilution beyond MVD, producing a negative result that could not have detected endotoxin at the product limit.
• Gel-clot tubes re-read outside the 60 minute window, or read early, until a preferred outcome appeared.
• A new lysate lot put into reportable use without confirming its labelled sensitivity, so the assay sensitivity for every run on that lot is unsupported.
• Standard curves accepted with a correlation coefficient below 0.980, or a curve carried over from a previous run.
• Endotoxin limits and MVD recomputed at the bench each time, with arithmetic errors reaching release decisions.
• Invalid runs deleted rather than retained, leaving a gap in the acquisition audit trail.
• A failing result retested into compliance before an OOS investigation was opened, or averaged with a passing replicate.
• Labware used from a depyrogenation cycle whose qualification had lapsed, or LAL reagent water with no lot recorded.

How to adapt this SOP

1. Set your document number, owner, and effective date in the header, and name the products and sample types in scope in section 2.
2. State in section 2.1 which compendial chapter applies per product, and confirm whether the method is registered in any marketing application before changing it.
3. Fill the reagent controls in section 5.1 with your actual labware source, depyrogenation cycle reference, and LAL reagent water handling limits.
4. Insert your validated pH range, routine dilutions, and plate map references in sections 5.5 and 5.6.
5. Point the cross-references in sections 2, 5.10, and 5.11 to your real method suitability, deviation, and OOS procedures.
6. Keep the numeric criteria in sections 5.7 and 6 as written unless your registered method specifies something tighter, and confirm each against the current compendial chapter before issue.
7. Where a product is at risk of low endotoxin recovery, link this procedure to the LER study for that product so analysts know the routine positive product control does not answer the masking question.