Endotoxin limits and product-contact/buffer controls in recent FDA BLA reviews

Two quality themes recur in FDA biologics review memos from the last five years (mid-2021 to mid-2026): how endotoxin (pyrogen) limits are set and justified, and how the agency scrutinizes the materials and solutions that touch the product, from single-use manufacturing components to the buffers and container-closure systems that hold the drug substance and drug product. The examples below are drawn from CBER-reviewed BLAs (cell and gene therapies, vaccines, plasma products) and CDER-reviewed BLAs (monoclonal antibodies, biosimilars, therapeutic proteins). Many of the exact numeric acceptance criteria are redacted as confidential commercial information in the released memos, but the review rationale is visible and consistent enough to describe the agency's expectations.

How FDA sets and justifies endotoxin limits

The dominant approach across cell and gene therapy BLAs is to anchor the endotoxin limit to the maximum patient dose and a per-kilogram exposure threshold rather than a fixed compendial number. FDA repeatedly describes the limit as a concentration derived so that a patient will not receive more than the allowed endotoxin exposure at the maximum administered dose.

For CARVYKTI (ciltacabtagene autoleucel), FDA stated the endotoxin limit concentration is based on a maximum human dose, ensuring a patient would not receive more than the specified endotoxin exposure ³⁴.
For AUCATZYL (obecabtagene autoleucel), the threshold was defined mathematically based on worst-case product characteristics and thresholds for endotoxin exposure ².
For LENMELDY (atidarsagene autotemcel), the criterion was chosen so a patient would not exceed a maximum exposure at a bounded administered volume, tying the limit to a dose/exposure calculation ¹.
For LANTIDRA (donislecel-jujn), FDA set the drug product endotoxin limit against compendial exposure limits for infusions or injections of one hour or less in duration ⁵⁶.

For AAV gene therapies the limit is translated through the nominal titer. In HEMGENIX (etranacogene dezaparvovec-drlb), FDA justified the drug product endotoxin specification by translating it against the nominal titer of 1E13 vg/mL and concluded the resulting dose-related exposure was acceptable; the specification-setting approach relied on late-phase and process-validation/PPQ manufacturing experience, with some assays carrying action limits and post-marketing reassessment commitments ⁸¹¹. BEQVEZ (fidanacogene elaparvovec-dzkt) lists endotoxin among the drug product release specifications but the released memo does not show the underlying calculation ⁹¹⁰.

On the protein/mAb side, CDER reviews are more explicit about units and thresholds even when the numbers are redacted:

MERILOG (insulin aspart-szjj): FDA described the endotoxin limit in EU/mL, consistent with the drug product release specification, below the USP monograph mg-based specification, and below the USP safety threshold expressed in EU/kg; the applicant committed to reassess the limit once sufficient manufacturing data accrue ²².
HYMPAVZI (marstacimab-hncq): a drug substance endotoxin acceptance criterion of no more than 2 EU/mL, with the drug product criterion described as equivalent to less than 0.25 EU/mg based on the target protein concentration ²⁵.

A practical cross-specification consistency point appeared in YUFLYMA (adalimumab-aaty): FDA flagged that an incoming-component endotoxin limit exceeded the release limit and asked the sponsor to correct it so a batch cannot pass the component limit yet fail release ¹⁶³⁴.

Units matter: volumetric (EU/mL) versus weight-based (EU/mg) limits

A consistent FDA position in recent liquid biologic reviews is that endotoxin limits for liquid drug substance, drug product, and in-process controls should be expressed volumetrically (EU/mL), not by protein mass (EU/mg). The reasoning is that endotoxin reflects microbial ingress and is independent of product concentration, so a volumetric limit is more stable batch to batch and avoids conversion variability.

RYZNEUTA (efbemalenograstim alfa-vuxw): FDA recommended reporting endotoxin volumetrically as EU/mL for liquid DS/DP in-process controls and release specifications; the sponsor agreed ¹³.
NYPOZI (filgrastim-txid): FDA made the same point, noting the assay output is in EU/mL and that EU/mg conversion adds variability, and also said the limit should be lowered based on process capability ¹⁷²¹.

FDA also pushes sponsors to tighten endotoxin limits once enough batches demonstrate process capability. In JUBBONTI/WYOST (denosumab-bbdz), FDA noted the limit was set on a limited dataset and asked the sponsor to re-evaluate and tighten it after 30 drug substance batches, with the updated limit to be submitted in a later Annual Report ⁹⁵⁹⁶.

Low endotoxin recovery (LER) and the rabbit pyrogen test

When the compendial bacterial endotoxins test (LAL) cannot reliably recover endotoxin, often because of the surfactant/chelator matrix in modern protein formulations, FDA has accepted the rabbit pyrogen test (RPT) as an interim release test and required post-marketing development of an LER-mitigating in vitro method.

OPDUALAG (nivolumab/relatlimab-rmbw): FDA found the endotoxin method could not reliably detect endotoxin due to LER, required RPT as an interim release test, and imposed a post-marketing commitment to develop an endotoxin method that mitigates LER (LER study on 3 lots, method qualification on 3 lots), with the validated method to replace RPT on supplement approval ³³.
ENTYVIO (vedolizumab): RPT would serve as the interim release test until a suitable in vitro endotoxin method is developed, with the endotoxin specification retained in 3.2.P.5.1; FDA noted RPT might be avoidable if spiked drug product could be shown non-pyrogenic in rabbits ⁹⁹.
XENPOZYME (olipudase alfa-rpcp): FDA required development of an LER-mitigating endotoxin method, method qualification on three lots, and LER study results on three lots, with RPT replaced upon supplement approval ¹⁰⁰.
IMAAVY (nipocalimab-aahu): FDA required a supplemental LER study examining hold-time effects on endotoxin recovery across three batches of each presentation, spiked with Reference Standard Endotoxin or Control Standard Endotoxin susceptible to LER, and investigation of alternative detection methods if LER is confirmed ¹⁰¹¹⁰².

The IMAAVY requirement highlights the hold-time dimension of LER: endotoxin recovery is tested after storage intervals because masking develops over time in the formulation matrix.

Product-contact materials: extractables/leachables and single-use systems

FDA reviews product-contact materials through a risk-based extractables and leachables (E&L) framework. Sponsors are expected to inventory every material contacting the product during manufacturing, score each by contact time, temperature, surface-area-to-volume ratio, solution properties, sterilization, and material resistance, then perform extractables and simulated/real-time leachables studies on the higher-risk items, with toxicological qualification against an analytical evaluation threshold (AET) and permitted daily exposure (PDE) or ICH M7/Q3C limits.

Cell and gene therapy BLAs, where the container closure is often a cryobag and the process is largely single-use:

LENMELDY (atidarsagene autotemcel): the process-wide E&L assessment flagged the DP cryobag and certain transfer/washing components as high risk and subjected them to extractable and simulated-leachable studies under representative and exaggerated conditions; FDA later found the initial container-closure leachables package insufficient to cover cumulative leachables across freezing, thawing, and clinical-site preparation and required additional work as a post-marketing commitment ⁷⁴⁸³.
TECELRA (afamitresgene autoleucel): FDA identified a high-risk leachables process step and said the accelerated-like study did not adequately cover all high-risk steps or real-time shelf life, requesting additional assessment including cumulative and elemental leachables; the highest E&L risk scores were tied to the DP container-closure system, and the cryogenic storage container was tested in an extractables simulation ⁷⁵⁸⁰⁸².
LANTIDRA (donislecel-jujn): FDA raised a deficiency on unassessed E&L from single-use consumables and sterilized reusable equipment; the sponsor's risk assessment concluded low leachables risk for each component ⁷²⁷³.
VYJUVEK (beremagene geperpavec-svdt): concerns on leachables/extractables for product-contact components, the container closure, and the administration device were resolved through information requests ⁸⁴⁸⁵.

Vaccines and prefilled-syringe products, where the closure system dominates the risk profile:

SPIKEVAX (COVID-19 mRNA vaccine): FDA reviewed E&L for the vial/stopper primary closure using worst-case extraction profiles from vendor materials data and toxicology assessment on calculated worst-case extractables per dose ⁷⁶.
CAPVAXIVE (21-valent pneumococcal conjugate vaccine): single-use product-contact materials were qualified by grouping polymers by material/manufacturer/sterilization/component type, selecting representatives, and scoring by route, proximity to final product, contact time, and surface-area-to-volume ratio; no high-risk PCMs were identified (four medium-risk) ⁷⁷⁷⁸.
VIMKUNYA (recombinant chikungunya vaccine): an elemental extractables study on the syringe, tip cap, and rubber plunger showed worst-case plunger compounds above PDE, but FDA judged clinically significant leaching unlikely given the plunger coating and aqueous formulation, with a shelf-life leachables study ongoing ⁸⁶⁸⁹.
MNEXSPIKE (COVID-19 mRNA vaccine): component biocompatibility and chemical characterization for the prefilled syringe, with barrel E&L data in 3.2.P.2.4 ⁸⁸.
TICOVAC: the borosilicate glass syringe, rubber plunger stopper, and tip cap were found appropriate, with container-closure integrity demonstrated by testing ⁹⁰.

CDER-reviewed mAbs and biosimilars show the same AET/PDE machinery applied rigorously to vials and prefilled syringes:

TYENNE (tocilizumab-aazg): the applicant identified 24 at-risk manufacturing-contact materials, tested blank buffer, materials, and samples stored at 5 ± 3°C by GC/MS and LC/MS with no non-volatile organics above 2x AET, and screened elemental impurities by ICP/MS below the 30% PDE control threshold. For the IV drug product in a glass vial with stopper and crimp seal, leachables on product and placebo filled with formulation buffer showed no relevant elemental impurities and organic leachables qualified against PDEs with no nonclinical safety concern ⁴²⁴⁴⁴⁷.
IDACIO (adalimumab-aacf): prefilled-syringe leachables were run with syringes stored horizontally to maximize contact with the rubber stopper (worst case); nine compounds were detected across long-term and stressed conditions, each below its established PDE ⁴³.
COSENTYX IV (secukinumab): E&L studies excluded the aluminum cap because it does not contact the drug product; AET was set using PQRI recommendations, and no leachables at or above AET were detected in vials stored up to 24 months, including filter extracts and most tubing/connectors/gaskets ⁴⁸⁴⁹.
TOFIDENCE (tocilizumab-bavi): FDA calculated an AET of 0.0375 µg/mL from a slope of contribution threshold of 1.5 µg/day; 7 chemicals exceeded AET in the stopper extraction and were tracked as potential leachables, and the worst-case 4 mL fill in a 6 mL vial showed no leachables above reporting limits. FDA accepted the toxicological risk assessment and did not request improved method sensitivity given the low-risk glass vial/rubber stopper system ⁶².
ENTYVIO SC (vedolizumab): a risk assessment using extraction capability, contact time, surface area, temperature, material resistance, and USP Class VI status identified the DP container and closure as higher-risk, triggering extractable studies under aggravated solvent/time/temperature conditions ⁵³.

Buffer, excipient, and component controls

Beyond E&L, FDA reviews the specifications on the buffers, excipients, and primary packaging components that define the product's contact environment.

HEMGENIX: FDA reviewed component specifications for vials (glass-container standards, supplier testing), stoppers (rubber-closure standards, incoming inspection, representative sterility/particulate/endotoxin testing), and seals (vendor certificate and biological-indicator sterility), later asking the applicant to determine sterility and endotoxin levels for the seals, which was addressed ⁷⁰.
PRIORIX: because residual neomycin sulfate appeared in growth/maintenance media and some bulk-production buffers but not in the harvest stabilizer or formulation media, FDA asked how it was controlled; the sponsor provided a theoretical worst-case estimate of residual neomycin sulfate per dose in the final container ⁹¹.
BALFAXAR (prothrombin complex concentrate, human-lans): reconstituted in water for injection, with sterility and endotoxin controls in the release context; FDA found the DS and DP specification acceptance criteria not adequately justified and asked for revised justification and supporting data ⁹³⁹⁴.

The BALFAXAR example is a useful reminder that specification justification, not just the presence of a test, is what FDA reviews: an endotoxin or bioburden control on a buffer or reconstitution solution must be supported by data and a documented rationale.

Practical takeaways for a BLA CMC dossier

Justify the endotoxin limit against the maximum patient dose and the EU/kg exposure threshold (the K/M approach), and reconcile it with the USP monograph and compendial exposure limits. For cell/gene therapies, tie it to the maximum administered dose or volume; for AAV, translate through the nominal titer ¹³⁵⁸.
Express endotoxin limits volumetrically (EU/mL) for liquid DS/DP and in-process controls, and be prepared to tighten the limit once process-capability data (often on the order of ~30 batches) accrue ¹³¹⁷⁹⁵.
Screen for low endotoxin recovery early. If LER is present, expect to defend an interim rabbit pyrogen test and commit to developing an LER-mitigating in vitro method, with hold-time-based LER studies across multiple lots ³³⁹⁹¹⁰⁰¹⁰¹.
Build a documented, risk-scored E&L program covering every product-contact material through the full process and shelf life, including cumulative and elemental leachables and real-time (not just accelerated) conditions; expect the container-closure system to be the highest-risk element and worst-case orientations/fills to be required ⁴²⁶²⁷⁴⁷⁵.
Support every buffer, excipient, and component control (bioburden, endotoxin, sterility, residual media components) with a justified acceptance criterion and consistency across the incoming-component and release specifications ¹⁶⁷⁰⁹¹⁹³.