Since it was first reported by Chen and Vinther in 2013 (1), the phenomenon known as low endotoxin recovery (LER) has been broadly observed in certain matrices commonly used for biologic formulations and certain therapeutic proteins. These observations have generated concerns that a pharmaceutical product contaminated with endotoxin may go undetected by the compendial USP <85>/EP 2.6.14./JP 4.01 bacterial endotoxin test (BET). In response to these reports, the U.S. FDA’s Center for Drug Evaluation and Research (CDER) recently began asking companies to conduct endotoxin spike/hold recovery studies to determine whether a given biological product is affected by LER (2–4). As a result, numerous spike/hold recovery studies have been carried out by many individual firms hoping to ameliorate FDA’s concern. These studies, however, have often produced confounding and sometimes contradictory results with regard to the cause, biochemical mechanism, and biological relevance of the LER phenomenon, likely due to variations in study designs, experimental procedures, and/or type of endotoxin used. Clearly, there is a mutual interest and desire for the industry and FDA to work together to develop a science-based and data-driven strategy in dealing with the LER phenomenon.

Recognizing the significance and complex nature of the LER issue, PDA’s Biotechnology Advisory Board (BioAB) sanctioned the LER Task Force in early 2015.This Task Force is composed of subject matter experts from academia, FDA, the biopharmaceutical industry, and reagent-supplier/testing companies. Many of the firms that have submitted LER data to the FDA are represented on the PDA LER Task Force, ensuring broad representation. The Task Force, however, will continue to reach out to maximize industrial participation.

The task force has three specific goals:

1. Investigate the root cause of LER
2. Standardize the experimental protocols for spike/hold recovery studies
3. Identify the potential safety impact of the LER phenomenon

As part of the effort to achieve these goals, the Task Force sponsored the first workshop exclusively dedicated to LER in March 2016 in San Antonio, coinciding with the 2016 PDA Annual Meeting. To facilitate the discussion, a questionnaire was sent out prior to the workshop to individual participating firms to collect specific information relevant to the LER phenomenon and spike/hold recovery studies. This approach proved to be highly valuable and effective since such details/specifics were usually not included or shared in conventional meetings.

Once at the two-and-one-half day workshop, attendees heard 11 presentations on endotoxin spike/hold recovery studies. Approximately two hours were allocated for each speaker, divided roughly between a 45 minute oral presentation and in-depth Q&A discussion. Highlights extracted from the workshop are summarized below.

Lack of Standardized LER Protocol Complicates BLA Review

Due to the lack of a standardized protocol, individual companies perform spike/hold recovery studies differently as reflected by the information provided to the FDA. This sometimes complicates the BLA review process by the agency and highlights the urgent need to establish a harmonized procedure for spike/hold recovery studies.

Although there was extensive discussion of how spike/hold recovery studies should be executed with regard to temperatures in the context of GMP manufacturing conditions, the issue remains unresolved at this time. It has been observed that different compendial bacterial endotoxin testing (BET) methods—kinetic turbidimetric assay (KTA), kinetic chromogenic assay (KCA), and gel clot assay (GCA)—may have different susceptibility to LER. FDA will accept change to another compendial method that does not show LER, provided that the results are consistent and reproducible. Change to a noncompendial method will even be accepted if adequate method validation data and relevant information are provided.

Several firms presented data from spike/hold recovery studies using both purified lipopolysaccharide (LPS) and in-house prepared natural occurring endotoxin (NOE). In some studies, LER was observed only when LPS was the spiking analyte, but not with NOE. In other studies, no difference was seen between LPS and NOE with regard to LER. In one study, it was shown that NOE prepared from different bacterial species exhibited a great degree of variability in recovery when spiked into a drug product formulated in a phosphate and PS80 matrix. The same study also reported that reference standard endotoxin had substantially slower rate of activity loss than control standard endotoxin in the drug product matrix. The exact bases for such confounding and even contradictory results remain unknown and the debate of using LPS versus NOE in spike/hold recovery studies is likely to continue for the foreseeable future.

Different Strokes for Different LERs

Data shared at the workshop showed that LER could be triggered by very different factors. Multiple companies reported that therapeutic products themselves could potentially cause LER. Although the combination of chelator/PS80 often leads to LER, it is not always the case, suggesting the involvement of other unknown elements or synergistic effects of individual components.

Participants also discussed that the role of surfactants (e.g., PS80) in LER is significantly less profound than chelators (e.g., citrate). One participant reported that LER observed in a DP formulated with citrate and PS80 could be overcome by the addition of divalent Mg++ prior to testing, suggesting that LER is readily reversible. A similar approach, however, did not result in successful rescue of recovery in other companies. The data and experimental details of these studies will be collected and analyzed in order to better understand the cause for the apparent discrepancy.

In addition, the Task Force learned that LER could be overcome by a “de-masking” process using proprietary reagents. However, the effectiveness of de-masking appears to be product/formulation dependent.

Finally, it was shown that LPS recovery can also be influenced by sampling scheme and vortex time (5).

Recommendations

Based on the data and experience shared at the workshop, the group has proposed several recommendations with the objective of ensuring the relevance of the spike/hold recovery studies and harmonizing the experimental procedure as much as possible:

1. If there is no LER observed in the final DP, it is not essential to perform the spike/hold recovery studies with the prior process intermediates such as drug substance (DS)
2. Spike/hold and recovery studies should be carried out in a manner representative of the corresponding QC BET testing with regard to materials, containers, and experimental procedures
3. LER is defined as the endotoxin activity falling below 50% of the spiked amount at two consecutive time points; and
4. Sampling repeatedly from a single spiked container should be avoided. Instead, reverse assay (spiking independently on different days and testing all the spiked samples on the same day) or multiple-aliquot approach (dispensing the spiked material into individual containers and testing individual containers on different days) is strongly recommended

Workshop Moves Debate Forward

The LER phenomenon has generated much discussion and debate in almost every microbiology conference and workshop during the last three years. Experimental results and data presented at this workshop indicate that while considerable progress has been made during this period, fundamental questions concerning the LER phenomenon remain unanswered.

All individual members of the PDA LER Task Force are committed to work together to advance our understanding of the LER phenomenon.

The LER Task Force currently has four subgroups with each working on a different aspect of the LER phenomenon. Subgroup 1 is focused on providing a clear guidance for spike/hold recovery studies. Subgroup 2 is working on understanding the underlying mechanism of the LER phenomenon and evaluating whether it is possible to develop a procedure/method to produce an endotoxin standard for spike/hold recovery studies. Subgroup 3 is concentrating on assessing the potential safety risk of the LER phenomenon by careful evaluation of the available data. And finally, Subgroup 4 is devoting all its efforts to providing clear definitions for terms relevant to the LER phenomenon using the USP Pharmacopeial Forum (PF) as a starting point.

There is no doubt that more research and investigation are needed in order to better understand the LER phenomenon, elucidate its underlying mechanism, and determine its potential clinical significance. The LER Task Force is confident, however, that the group’s specific goals will be achieved by working together as a team and adhering to data-driven/ science-based principles.

[Editor’s Note: Learn more about the Task Force’s workshop in session “A2: Challenges in Endotoxin Recovery,” Oct. 24, 1:30 p.m. at PDA’s pharmaceutical microbiology conference.”]

The task force acknowledges the following individuals for sharing their information and data at the workshop: Mazukazu Tsuchiya, Charles River; John Dubczak, Charles River; Jay Bolden, Eli Lilly; Stefan Ishak, Sandoz; Johannes Reich, Hyglos; Phil Villari, Merck Sharp & Dohme; Cheryl Platco, Merck; Anders Thorn, Novo Nordisk; Chris Knutsen, BMS; Ned Mozier, Pfizer; and Karen McCullough; MMI Associates.

References

1. Chen, J., and Vinther, A. “Low Endotoxin Recovery in Common Biologics Products.” Presented at the 2013 PDA Annual Meeting, Orlando, FL. April 2013.
2. Guidance for Industry Pyrogen and Endotoxins Testing: Questions and Answers, U.S. Food and Drug Administration, June 2012
3. Hughes, P. “Endotoxin – A FDA Perspective.” Presented at the PDA 10th Annual Global Conference on Pharmaceutical Microbiology, Bethesda, MD, October 2015.
4. Hughes, P., et al. “Low endotoxin recovery: An FDA perspective.” BioPharma Asia 4 (2015): 14–25
5. Bolden, J. et al., “Endotoxin recovery using Limulus amoebocyte lysate (LAL) assay.” Biologicals. 16 (2016): S1045– 1056.