Protocol: Low Endotoxin Recovery (LER) Hold-Time Study

This is a ready-to-use protocol for a low endotoxin recovery (LER) hold-time study, also called an endotoxin masking study. Replace every <<FILL: ...>> placeholder with your own specifics, set your document numbers and dates, attach the reagent certificates and raw assay records, and route it through your normal document control, review, and approval. A worked filled specimen follows the template. This is educational reference content, not regulatory advice; adapt it to your own product and confirm the current expectations in each market where the product is or will be registered.

Approval page

1. Objective

To determine, with documented evidence, whether the finished product formulation named on the approval page causes a progressive loss of detectable bacterial endotoxin activity over time when undiluted product is spiked with a known amount of endotoxin and held under process-relevant conditions, and to establish whether the routine bacterial endotoxin test can be relied on to detect endotoxin in that matrix.

The study is a characterisation of the method in the matrix. It is not a test of the batch and it does not generate a release result.

2. Scope

This protocol covers the undiluted spike-recovery hold-time study for the finished product as administered to the patient, at the conditions and time points defined in section 8.

In scope:

• Finished product at label concentration, undiluted and unfrozen.
• The compendial endotoxin method used for routine release of this product.
• A risk assessment of whether the finished product study represents the active substance.

Out of scope:

• Routine batch endotoxin testing, governed by the bacterial endotoxin testing SOP.
• Derivation of the endotoxin limit and MVD, held on the endotoxin limit and MVD calculation worksheet.
• Product-specific method suitability (interference) validation at the routine test dilution, which is a different study answering a different question (see section 4.3).
• QC sample hold time, which is also a different study (see section 4.4).

3. Background

Low endotoxin recovery is the progressive loss of detectable endotoxin activity when a known endotoxin spike sits in undiluted product over time. The endotoxin remains physically present; the assay stops seeing it. The working mechanistic understanding is that certain formulations disassemble the supramolecular aggregates that lipopolysaccharide forms in solution, and the amebocyte lysate cascade depends on that aggregation state to respond.

The practical consequence for a release laboratory is uncomfortable: a routine bacterial endotoxin test can return a clean result on a genuinely contaminated sample, and the routine controls will not warn you, because those controls are run at a dilution where the masking no longer operates.

Regulatory attention to this has come principally from the European Medicines Agency. EMA’s questions and answers for biological medicinal products carries the questions on low endotoxin recovery and endotoxin masking, and that document was updated on 20 July 2026, with entries on the endotoxin LAL test alongside bioburden testing, pre-use filter integrity testing, and viral filtration. Because a questions-and-answers page is revised without the version history a guideline carries, read the current published version for the operative wording before relying on any summary of it, including this one.

For study design methodology, PDA Technical Report No. 82, Low Endotoxin Recovery, addresses the topic specifically and is recognised as a relevant design reference. Its content is not reproduced here. Consult the technical report directly for its methodology rather than working from any secondary description of it.

Background reading: bioburden and bacterial endotoxin testing.

4. Four things this protocol is designed to get right

4.1 The spike goes into undiluted product

This is the whole point of the study, and it is the step most often done wrong. Masking is disrupted by dilution. A spike added to product that has already been diluted to the routine test concentration will usually be recovered, because the formulation is no longer concentrated enough to act on the endotoxin. Spiking undiluted product is what makes the study capable of detecting masking at all.

4.2 The routine positive product control does not answer this question

The positive product control run with every routine assay is a single time point spike at the working dilution. It answers whether the diluted matrix inhibits or enhances the reaction at the moment of testing. It cannot answer whether the undiluted formulation progressively masks endotoxin over hours or days, because neither the concentration nor the time is right. A product can pass its positive product control on every batch for years and still be strongly LER-positive. Treating a passing positive product control as evidence against LER is a category error and a recurring finding.

4.3 Method suitability and LER are different studies

Method suitability (the inhibition and enhancement test) qualifies the method at the routine dilution and is required before any result is reportable. The LER study characterises the undiluted matrix over time. Passing one says nothing about the other. Both are needed for an at-risk formulation.

4.4 LER hold time is not QC sample hold time

These are routinely confused and they answer opposite questions.

Establishing one does not establish the other. Do not cite a QC sample hold-time study as an LER assessment, or the reverse.

5. Risk trigger: is this study required for this product?

Complete this assessment and retain it whether the answer is yes or no. A documented “not at risk” conclusion is itself part of the control strategy.

Trigger rule. A formulation combining a surfactant with a chelating species is treated as at risk and this study is performed. Because that description covers a large share of monoclonal antibody and recombinant protein formulations, the study is the expectation for those products rather than the exception. Where the risk factors are absent, record the assessment and the conclusion, and revisit it at any formulation change.

Note that LER is a property of the formulation, not of the contamination. A well controlled process with no endotoxin history can still be LER-susceptible, and the susceptibility has to be characterised rather than argued away on the basis of a clean manufacturing record.

6. Prerequisites (verify and record before execution)

7. Roles and responsibilities

8. Study design

8.1 Test articles and spiking

The spike volume matters. Adding a large volume of aqueous spike dilutes the formulation and can suppress the very effect the study is looking for. Keep the spike volume small and record it.

Consider including a naturally occurring endotoxin preparation alongside the purified control standard endotoxin, because masking behaviour can differ between them. Where only the control standard is used, record the justification.

8.2 Hold conditions

Conditions must reflect what actually happens to the product, not an arbitrary storage condition. Derive them from the process and the product handling.

The longest duration studied should cover the longest realistic manufacturing hold for that material, plus a margin. A study that stops at 24 hours does not support a 14 day bulk hold.

8.3 Time points

A minimum of four time points is evaluated per condition, so the trajectory can be seen rather than a single end value. Add points where the process has a defined hold.

Record the actual time of each test, not only the nominal interval. A point tested a day late changes the interpretation of the curve.

8.4 Controls at every time point

If the positive water control also loses recovery, the loss is not attributable to the formulation and the study cannot conclude LER until that is resolved.

8.5 Testing and calculation

1. At each time point, withdraw an aliquot from the held vessel, dilute to a dilution at or below MVD using the routine method, and assay by the routine compendial method.

2. Correct the measured value for the dilution factor.

3. Calculate recovery:

   Recovery (percent) = (dilution-corrected measured endotoxin in spiked product) / (nominal spike concentration) x 100

4. Record recovery for every replicate, every condition, and every time point. Do not average across conditions.

5. Plot recovery against time for each condition.

9. Acceptance criteria

Two categories, and they must not be confused. The validity criteria decide whether the study can be interpreted. The decision rule decides what the study concluded.

9.1 Study validity criteria

9.2 LER decision rule

Two consecutive time points at which recovery falls below 50 percent is treated as demonstrating low endotoxin recovery for that condition.

Fifty percent is the lower end of the recovery range the compendial methods accept, so a sustained fall below it means the method is no longer recovering the spike. Requiring two consecutive points rather than one prevents a single aberrant assay from driving the conclusion.

10. Test cases

Execute in order. Record actual results and initial each line.

TC-1 Prerequisites and system readiness

TC-2 Spike preparation and T0 recovery

TC-3 Hold and time-point testing, condition C1

TC-4 Hold and time-point testing, condition C2

TC-5 Additional conditions (where defined)

TC-6 Data evaluation and LER determination

TC-7 Active substance representativeness assessment

The investigation focuses on the finished product as administered to the patient. Where the active substance matrix differs, do not assume the finished product study covers it; assess it and record the reasoning either way.

11. Deviation handling

Any departure from this protocol (a missed or late time point, an invalid assay, a storage excursion, a spike concentration outside the nominal, a container substitution) is recorded as a protocol deviation, assessed for impact on the conclusion, and dispositioned by QA per <<FILL: deviation SOP>> before the affected data is used.

Before concluding LER from a losing curve, rule out the alternatives in this order: assay validity, positive water control behaviour, container adsorption, spike preparation error, and storage excursion. A conclusion of LER should be what remains after those are excluded, not the first explanation reached.

12. If LER is demonstrated: response options

Finding LER is not a product failure and it is not a batch event. It is a method limitation that has to be addressed before the method can be relied on.

1. Optimise the compendial method or develop an alternative for this matrix. Approaches include demasking or dispersing pretreatments, alternative diluents, adjusted sample preparation, and reagent systems that behave differently in the matrix. Any optimised or alternative method is validated for the matrix and routed through change control, and where the method is registered the change is a filing action in each affected market.
2. Set the finished product endotoxin specification as low as reasonably achievable, on the basis of actual manufacturing data rather than a round number. Where the assay’s sensitivity in that matrix is partly compromised, a tighter specification restores part of the safety margin that masking removed.
3. Reinforce upstream endotoxin control, since the assay is a less reliable last line of defence in this matrix: water system control, component and container depyrogenation, bioburden control at the sterile filtration control point, and hold time control.
4. Place the data where reviewers expect it. The study data belongs in CTD section 3.2.P.5.3 (validation of analytical procedures) or 3.2.P.2.3 (manufacturing process development), with a cross-reference from the other, and in Module 3 for a new marketing authorisation application. Confirm the placement with Regulatory CMC.
5. Record the conclusion in the contamination control strategy so the limitation is visible to the people who rely on the endotoxin result. See Annex 1 contamination control strategy.

13. Conclusion

14. References

EMA, questions and answers for biological medicinal products (the entries on low endotoxin recovery and endotoxin masking), updated 20 July 2026. Read the current published version. PDA Technical Report No. 82, Low Endotoxin Recovery. Cited as a study design reference only; consult the technical report itself. USP <85> Bacterial Endotoxins Test; USP <86> Bacterial Endotoxins Test Using Recombinant Reagents (official 1 May 2025); USP <1085> Guidelines on the Endotoxins Test. 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 of the consolidated arrangement from 1 January 2027. Cite 2.6.14 Method G rather than 2.6.32 in any document that will still be in force after that date. ICH Q2(R2) Validation of Analytical Procedures; ICH Q9 Quality Risk Management. EU GMP Annex 1, Manufacture of Sterile Medicinal Products.

Confirm the current version of each reference before issue. The compendial chapters and the technical report are copyrighted; their content is not reproduced here.

15. Attachments

Risk trigger assessment; CSE certificate and traceability; lysate lot qualification; LAL reagent water release record; storage unit monitoring records for each condition; raw assay records and standard curves for every time point; recovery calculation sheets with second-person verification; deviation reports; container and labware depyrogenation records.

16. Revision history

Filled specimen

The following shows key sections completed for an example monoclonal antibody drug product. Company, product, and numbers are illustrative; replace them with your own.

Product: MAB-2210 drug product, 25 mg/mL, in 20 mM histidine with 10 mM sodium citrate, 0.02 percent polysorbate 80, pH 6.0, presented in 10 mL vials for intravenous infusion after dilution.

Risk trigger: Surfactant present (polysorbate 80) and two chelating species present (citrate, histidine). Bulk drug product is held at 2 to 8 C for up to 14 days before fill, and compounded product stands at room temperature during in-use handling. Study required.

Design: Three batches. Spike 50 EU/mL from CSE lot 887 into undiluted product in depyrogenated 10 mL borosilicate vials, spike volume 1.0 percent of the test article volume. Conditions C1 = 2 to 8 C and C2 = 20 to 25 C. Five time points per condition: T0, 24 hours, 3 days, 7 days, 14 days, with 28 days added at C1 to cover the bulk hold plus margin. Kinetic chromogenic assay, lambda 0.005 EU/mL, all assays run at 1:100, well inside the MVD of 2500.

Results, condition C2 (20 to 25 C), batch A:

Results, condition C1 (2 to 8 C), batch A:

Controls: Positive water control at 20 to 25 C recovered 96 percent at T0 and 88 percent at 14 days, staying inside 50 to 200 percent throughout, so the loss in product is attributable to the matrix and not to spike degradation or vial adsorption. Unspiked product background was below the assay sensitivity at every time point. All runs met the routine assay controls, with standard curve correlation coefficients between 0.995 and 0.999 and positive product control recoveries between 84 and 108 percent at the 1:100 test dilution. Batches B and C followed the same pattern within a few percentage points.

Determination: At C2, recovery fell below 50 percent at 3 days and 7 days, two consecutive points, so LER is demonstrated at room temperature. At C1, recovery fell below 50 percent at 14 days and 28 days, two consecutive points, so LER is demonstrated at 2 to 8 C as well, developing more slowly. The 14 day bulk hold sits exactly at the point where recovery crosses the threshold, which is the operationally important finding.

Note on what the routine controls showed. Every positive product control in this study passed comfortably at the 1:100 test dilution, on the same days that the undiluted product was recovering 6 percent of its spike. That contrast is the reason this study exists and the reason a passing positive product control is not evidence against LER.

Active substance assessment: The active substance is formulated in the same histidine and citrate buffer with polysorbate 80 at a higher protein concentration. The matrices were judged similar enough that the finished product study is representative, and the reasoning was recorded rather than assumed. Had the active substance been held in a different buffer without surfactant, a separate study would have been raised.

Actions taken: Method optimisation was initiated and a dispersing pretreatment restored recovery to 78 percent at 14 days under condition C1; the optimised preparation was validated for the matrix and routed through change control, with a regulatory assessment of the filing impact in each market. In parallel the finished product endotoxin specification was tightened from the calculated limit of 12.5 EU/mL to 2.0 EU/mL on the basis of 36 months of manufacturing data in which no batch exceeded 0.4 EU/mL. The bulk hold was capped at 7 days pending the optimised method. The study data was placed in CTD section 3.2.P.5.3 with a cross-reference from 3.2.P.2.3, and the limitation was recorded in the contamination control strategy.

Common inspection findings this protocol prevents

• No LER assessment at all for a formulation combining a surfactant with a chelator, so the ability of the release method to detect endotoxin in that matrix was never established.
• A passing routine positive product control offered as evidence that the product does not mask endotoxin, when the control is run at a dilution where masking does not operate.
• A spike added to already-diluted product, producing a study that could not have detected masking.
• A single end-point spike recovery result presented as a hold-time study, with no trajectory and fewer than four time points.
• Hold conditions and durations chosen for laboratory convenience rather than derived from actual process holds, so the study does not cover the hold time in use.
• A QC sample hold-time study cited as an LER assessment, or the reverse.
• No positive water control, so a loss of recovery cannot be attributed to the matrix rather than to spike degradation or container adsorption.
• LER demonstrated and then closed with no method action and no specification review.
• The active substance assumed to be covered by the finished product study with no comparison of the two matrices.
• The study performed but the data not placed in the submission, so a reviewer cannot find it.

How to adapt this protocol

1. Complete the section 5 risk trigger first and retain it whether the answer is yes or no; a documented negative conclusion is part of the control strategy.
2. Derive the hold conditions in section 8.2 from your actual process holds and product handling, and extend the longest time point past the longest real hold.
3. Set the spike concentration and, importantly, the spike volume fraction, so the spike itself does not dilute the formulation.
4. Decide and record whether a naturally occurring endotoxin preparation is included alongside the control standard endotoxin.
5. Keep the decision rule in section 9.2 as written unless you have a documented, scientific basis for something different, and keep the validity criteria separate from the decision rule so an invalid study is never reported as a negative finding.
6. Point the deviation cross-reference in section 11 to your real procedure.
7. Agree the CTD placement with Regulatory CMC before the report is written, not after.
8. Confirm the current EMA position and the current compendial text before issue, and consult PDA Technical Report No. 82 directly for design methodology.