Abstract
Recombinant Factor C (rFC) is non-animal-derived reagent used to detect bacterial endotoxins in pharmaceutical products. Despite the fact that the reagent was first commercially available nearly 15 years ago, the broad use of rFC in pharmaceutical industry has long been lagging, presumably due to historical single-source supplier concerns and the lack of inclusion in worldwide pharmacopeias. Commercial rFC reagents are now available from multiple manufacturers, thus single sourcing is no longer an issue. We report here the successful validation of several pharmaceutical products by an end-point florescence-based endotoxin method using the rFC reagent. The method is equivalent or superior to the compendia bacterial endotoxins test method. Based on the comparability data and extenuating circumstances, the incorporation of the end point fluorescence technique and rFC reagent in global compendia bacterial endotoxins test chapters is desired and warranted.
LAY ABSTRACT: Public health has been protected for over 30 years with the use of a purified blood product of the horseshoe crab, limulus amebocyte lysate. More recently, this blood product can be produced in biotech manufacturing processes, which reduces potential impacts to the horseshoe crab and related species dependent upon the crab, for example, migrating shorebirds. The pharmaceutical industry has been slow to adopt the use of this reagent, Recombinant Factor C (rFC), for various reasons. We evaluated the use of rFC across many pharmaceutical products, and in other feasibility demonstration experiments, and found rFC to be a suitable alternative to the animal-derived limulus amebocyte lysate. Incorporation of rFC and its analytical method into national testing standards would provide an equivalent or better test while continuing to maintain patient safety for those who depend on medicines and while securing pharmaceutical supply chains. In addition, widespread use of this method would benefit existing animal conservation efforts.
Introduction
Injectable pharmaceutical products are required to be sterile and pyrogen-free within a product specification (1). Animal-derived Limulus amebocyte lysate (LAL) reagent has been widely used for the detection of bacterial endotoxins in the pharmaceutical industry since the early 1980s. Ding and Ho first described the potential use of Recombinant Factor C (rFC) protein in place of LAL for endotoxin detection in 1997 (2). However, the pharmaceutical industry has been slow to adopt the rFC reagent for reasons such as lack of method description in any of the major pharmacopeias (e.g., USP, Ph.Eur., and JP) (3⇓–5) and a concern about the single-source supply.
rFC and rLAL products are now available from multiple manufacturers/suppliers, and recently, regulatory authorities in the United States and Europe have endorsed the use of the rFC reagent (4, 6). Currently, firms choosing to use a non-compendia method in lieu of a compendia method must go through an appropriate method validation process and demonstrate that the new method is equivalent or superior to the compendia one (3⇓–5).
Replacing the animal-derived LAL with rFC for endotoxin testing is consistent with the 3 R principles (replacement, reduction, refinement) for more ethical and sustainable use of animals for testing (7, 8). As part of Lilly's commitment to animal welfare and conservation, we have systematically evaluated the feasibility of using rFC to replace LAL for endotoxin testing. This article describes the results of various studies comparing the assay specificity, sensitivity, and robustness between rFC and LAL, as well as the validation approach used to demonstrate the comparability of rFC for endotoxin testing under good manufacturing practice (GMP) pharmaceutical manufacturing conditions.
Materials/Methods
LAL and rFC reagents, purified endotoxin standards, and LAL reagent water (LRW) were purchased from Lonza (Walkersville, MD) or Biomerieux (Bernried, Munich). The lyophilized LAL reagent was reconstituted with LRW prior to use. The rFC reagent was prepared by mixing a three-part liquid combination of a fluorescent substrate, buffer, and enzyme in a supplier-specified ratio prior to use. Beta glucan blocking buffer and beta glucan detection cartridges were provided by Charles River (Charleston, SC).
Fluorescence and absorbance readers used in the study included a BioTek (Winooski, VT) FLx800, a modified BioTek H1 (PyroWave XM, Lonza Walkersville, MD) and a BioTek ELx808 coupled with WinKQCL software (Lonza Walkersville, MD). An Endosafe PTS (PTS) hand-held absorbance reader was utilized for beta glucan detection (Charles River, Charleston, SC).
Proprietary drug products and drug substances, and related excipients, were obtained within Eli Lilly (Indianapolis, IN).
For the rFC-based tests, fluorescence measurements were carried out using 380/20 nm excitation and at 440/30 nm emission detection/bandpass. End-point reads occurred at the initial time point and after incubating for 60 or 90 min at 37 °C, following an initial 10 min incubation at 37 °C. Standard and sample results in raw fluorescence units were log-transformed to endotoxin units (EU) per milliliter and plotted. Data suitability was demonstrated by standard curve linearity, slope, y-intercept, replicate consistency (percent coefficient of variation, %CV), and positive and negative controls.
For the LAL-based tests, absorbance measurements were carried out via detection of a chromophore at 405 nm using optical density (OD) cutoffs. The sample was mixed with equal parts LAL, incubated at 37 °C, and kinetically assayed as described by the specific reagent supplier.
Robustness
Various test conditions were intentionally challenged to evaluate the robustness of the method. Acceptance criteria were set as recommended from the rFC supplier. The following parameters were evaluated in various combinations:
Sensitivity Setting: nominal value, ± 3. Because sensitivity setting is specific to a fluorescent reader and rFC kit, it must be established prior to use to optimize the detection range.
Sample/Reagent Volume: 100 μL ± 10% (100 μL of both sample and reagent ± 10%). The sample/reagent volume was altered to simulate an out-of-tolerance range pipette.
Incubation Time: 10 min, +10 min, +20 min. Incubation times were extended to simulate scenarios in which the reagent could not be immediately added to the 96 well plate.
Time to Reagent Addition: 0 min, +10 min, and +20 min at room temperature, light-protected. Reagent addition times were prolonged to mimic situations during which the reagent could not be immediately added to the 96 well plate following addition of the sample(s).
Validation
Product-specific method validations were performed by evaluating recovery of the purified endotoxin analyte in the presence of product at the approximate mid-point of the standard curve and in the upper and lower portions of the standard curve. Regulatory and compendia guidance allows the use of non-compendial alternate methods provided that scientific justification is established (3⇓⇓–6). Accuracy, precision, and the traditional compendia validation parameters such as inhibition/enhancement and pH suitability were assessed for each product. Linearity, range, specificity, limit of quantitation (LOQ) are not product-specific for the reagent. Data are typically generated on each routine test analysis against a standard curve with acceptance criteria that continually demonstrate laboratory operation in a validated state. Furthermore, data provided by Loverock et al. (9) was leveraged for exclusion of routine linearity, range, specificity, and LOQ product-specific validation. Robustness was performed as a one-time study as already described.
(a–c). The comparability of rFC and LAL to detect endotoxin from three organisms inoculated in two biopharmaceutical process buffers over 21 days.
Additional Specificity
The dynamic contamination simulation employed frozen bacterial cultures of Variovorax paradoxus (ATCC 17713), Delftia acidovorans (ATCC 15668), and Ralstonia pickettii (ATCC 27511). Each thawed bacterial culture was spiked into two common biopharmaceutical buffer matrices (10 mM histidine and 150 mM sodium chloride, pH 6.0; and 10 mM sodium citrate, 0.05% polysorbate 80, pH 6.0) and a water control for evaluation of the bacterial growth and associated endotoxin activity. The spiked buffers and control, in individual containers, were incubated at ambient temperature and sampled for bacterial enumeration after initial inoculation, daily for up to 4 days, and after 21 days. Sample endotoxin activity was quantified after approximately 24 hours, daily through 4 days, and after 21 days post-inoculation by diluting in LRW and using both the rFC and LAL assays. Samples were also tested for total aerobic microbial count consistent with the USP chapter 61 pour plate method (3) to obtain a corresponding result in colony-forming units (CFU) per milliliter.
Discussion/Conclusions
The results of the robustness study indicate the end-point fluorescence technique using rFC reagents is unaffected by certain anticipated lab excursions from the routine method within the studied ranges. All study parameters met pre-established acceptance criteria despite the various combination of method excursions at nominal or extremes (Table I). Such a robustness study could be utilized to evaluate the quality impact of laboratory events. This includes pipettes found to be out of tolerance, as the data indicate that “as found” standard criteria (e.g., International Organization for Standardization <1.5%), while desirable, may be conservative with respect to the intended purpose of the bacterial endotoxins test.
Results of the rFC Robustness Study
To date, we have successfully validated ten drug products, six drug substances, two pharmaceutical formulation excipients and pharmaceutical production–grade water (Table II). rFC comparability to LAL is demonstrated with and without the presence of endotoxin in the sample matrix (Table III). The results demonstrate that the two reagents are practically equivalent with respect to the reported results in unspiked samples. Due to different assay interference profiles, the result generated by the rFC could be slightly more or less sensitive than LAL. However, all results are well within assay specification, as products are not typically evaluated using a limit test. The positive control recovery results generated using rFC are equivalent to or superior to those generated using LAL. In the presence of endotoxin, the rFC demonstrated comparable results (50–200%) with a better accuracy range (74–151% LAL vs. 88–117% rFC) and better intra-assay precision (5–39% LAL and 0–11% rFC). We validated several pharmaceutical-use water grades using rFC reagent from two different suppliers and found them to be practically equivalent.
Results of the Product-specific Method Validations Using rFC Reagent for the Detection of Bacterial Endotoxins in Various Pharmaceutical Products
The Comparability of rFC and LAL in the Absence and Presence of Endotoxin Analyte across 18 Pharmaceutical Products and Excipients
Preclusion of the false-positive Factor G activation pathway using rFC is demonstrated from in-process drug substance samples that atypically exhibited endotoxin activity above the detection limit, but within product limits (Table IV). Cellulosic filtration material, a known β-1,4-glucan source, was used during the drug substance manufacturing process. The results were repeated and confirmed in a second laboratory using the same methodology. The samples were then diluted in beta glucan blocking buffer and shown to be free of endotoxin at the detection limit using the bacterial endotoxins test (BET) kinetic chromogenic technique. Orthogonally, the samples were tested using rFC and end-point fluorescence, which confirmed the samples to be free of endotoxin at the detection limit. Subsequent analysis for beta glucan indicated elevated levels relative to the negative LRW control. In a second incident with a different product, similar results were obtained. Therefore, rFC was shown to be more specific for endotoxin, or lack thereof, in the presence of suspected beta glucans, causing a false positive in the BET.
Two Examples of Potential LAL False Positives in Biopharmaceutical Drug Substance Manufacturing and the Insensitivity of rFC toward Suspected Beta Glucans from Cellulosic Filter Material
LAL and rFC activities compared favorably in their ability to detect endotoxin from actively growing Gram-negative bacteria in two common biotech buffer matrices—a 10 mM histidine, 150 mM sodium chloride, pH 6.0 buffer; and a 10 mM citrate, 0.5% polysorbate 80, pH 6.0 buffer—and an LRW control. Variovorax and Delftia species proliferated in all three matrices over the study time period, and their respective endotoxin activities were detected trailing 2–4 logs from the bacterial enumeration counts. The Ralstonia species failed to propagate in all three matrices upon inoculation, presumably due to unfavorable osmotic pressure conditions; however, the endotoxin associated with the bacterial skeletons was reliably detected.
The results of these studies indicate that the use of rFC is robust as an LAL replacement for BET and can be validated for the detection of bacterial endotoxins in a variety of pharmaceutical products. Furthermore, rFC was shown to be insensitive to beta glucan, thus minimizing false-positive results and improving assay specificity. Specificity for bacterial endotoxins was shown to be comparable under simulated contamination events. Additional studies and disclosure of product-specific method validations by multiple rFC reagent suppliers would further support industry adoption of the recombinant reagent. Adoption of the end-point fluorescence method using rFC reagent by the global pharmacopoeias will proactively allow broader, more proactive adoption for a non-animal-sourced reagent.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgments
The authors thank Drs. Dayue Chen, Aarron Garrett, and Amy Barker for their review of the manuscript.
- © PDA, Inc. 2017
Jump to section
Related Articles
Cited By...
- Recombinant Factor C Validation--Simpler Than You Think!
- Currently Available Recombinant Alternatives to Horseshoe Crab Blood Lysates: Are They Comparable for the Detection of Environmental Bacterial Endotoxins? A Review
- Comparison of Limulus Amoebocyte Lysate and Recombinant Factor C Assays for Endotoxin Detection in Four Human Vaccines with Complex Matrices