Evelyn Der, Senior Scientist, Roche Genentech, QC Technology Innovation and Implementation Group, South San Francisco, CA, USA.
Carmen Marin, QC Scientist Microbiology, F. Hoffmann La Roche Ltd, Kaiseraugst, Switzerland.
Viviane Grunert da Fonseca, PhD, Non-clinical Statistician, Roche, MSAT (Manufacturing, Science and Technology), Penzberg, Germany.
Lindsey Silva, PhD, QC Director, Roche Genentech, Microbiology, South San Francisco, CA USA.
Introduction
This paper presents Roche’s validation strategy of the recombinant factor C (rFC) endotoxin testing method. The validation approach uses statistical non-inferiority hypothesis testing based on spiked samples with different concentrations of commercial endotoxin standard. This approach is different from published literature studies while in accordance with the current regulatory framework and streamlines laboratory work. The validation activities are ongoing at the time of this publication at a Roche pilot laboratory. As the validation progresses, validation activities are to be continued at the other Roche global network QC operations and applied for water and product endotoxin testing.
Importance of a Sustainable Alternative Method at Roche
Roche sponsors its corporate 3Rs Incentive Awards, aiming to increase awareness and promote the dialogue among employees and scientists about the 3-Rs for animal welfare (refine, reduce, & replace). This program supports Roche’s mission to support innovation and advancement of animal welfare standards globally and replaces non-sustainable animal dependent analytical methods with alternative methods.
EU Directive 2010/63/EU adopted in 2010 by the Council of European Convention, replaced an earlier Directive from 2018 and harmonized standards across the EU. Based on the 3-R principles, this directive increases awareness of animal protection and welfare. Its ultimate goal is to replace animal use for scientific purposes (e.g., applied research, regulatory testing and production, education, and training).
At present, Roche tests globally over 39,000 water samples annually for endotoxins to monitor the quality of the water systems. Testing is dependent upon the compendia-based Limulus amebocyte lysate (LAL) test. There would be a significant impact on animal protection using an alternative method.
Proof of Concept (POC) Studies
Prior to the pilot phase of the rFC project, Roche performed preliminary proof of concept studies.
POC: Vendor Reagent Comparison
Different rFC vendor kits and respective readers were compared to a LAL kit and reader, and analyst to analyst variability was assessed. Included was an evaluation of analyst-to-analyst variability. Expected recovery was 50 to 200% for the spiked Reference Standard Endotoxin (RSE) recovery to rule out endotoxin inhibition and enhancement. Endotoxin recoveries of the spiked samples were compared against a theoretical spike of 5 EU/mL. For monoclonal antibody (mAB) #1, a lower than expected 47% RSE recovery observed with a vendor rFC kit and plate reader was likely equipment related. The Positive Product Control (PPC) and spiked RSE recovery results from the same and remaining vendors were acceptable, indicating the potential suitability of the rFC method. Refer to Table 1A.
POC: Interchange of rFC Reagents with Different Plate Readers
Evaluated was the interchange of different vendor rFC reagents and plate readers. Similarly, to the previous POC study, lower than expected spiked RSE recoveries using Vendor C’s plate reader and Vendor B’s reagents were attained. This event similarly was likely due to the initial reader setup. The remaining results were acceptable for %PPC and spiked RSE recoveries. Refer to Table 1B.
POC: Spiked Endotoxin Recovery Over Time
A mAb and LAL water control were spiked with 2 EU/mL RSE to evaluate possible storage-related masking effects.
The spiked test sample and control were held at 20-25˚C and tested at Time 0, Day 1, and Day 5. A comparison was made between the endotoxin recoveries from the test and control samples at each time point. Endotoxin masking was not evident. The data indicates that the rFC method is suitable for bacterial endotoxin testing in a drug product. Refer to Table 1C.
Regulatory Considerations
In Europe and as of January 2021, the rFC method is now an alternative pharmacopeial method, Ph Eur 2.6.32. For the bacterial endotoxins test, the monographs still reference Ph Eur chapter 2.6.14. Therefore, method 2.6.32 is considered an alternative to method 2.6.14 as described in the General Notices. Replacing a prescribed monograph method with another described in the Ph Eur does not require re-validation per se, other than in consideration of its use for a specific substance or product.(16) To replace the Ph Eur method 2.6.14 with 2.6.32, a reference to 2.6.32 should be added to the scope of these monographs as a supplemental or replacement bacterial endotoxin method.
Outside of Europe, the rFC method is an alternative non-pharmacopoeia method. In the US, rFC is considered an alternative test in current USP <1085> since it is not a method in USP <85>. A 2019 USP<85> draft added the rFC method; however, this rFC addition was cancelled, resulting in a draft standalone chapter <1085.1>. Therefore, the rFC remains an alternative method in the USP.
The rFC method is considered an alternative method in the ChP 9251 (informational) and the 18th edition of the JP, recently published.
Some FDA Office of Pharmaceutical Quality (OPQ) reviewers were either reluctant to accept BP/EP/JP quality standards or informed companies that USP/NF standards must be used as specifications even though the alternative standards are appropriate. The FDA’s Office of Pharmaceutical Quality MAPP (Manual of Policies and Procedure) 5310.7 clarifies for reviewers the appropriate use of quality standards found in alternative compendia. MAPP 5310.7 offers NDA/IND/ANDA sponsors the option of using standards from the alternative compendia provided that the rigor and acceptance criteria are equivalent to or better than USP/NF monograph requirements and the alternative analytical procedures are based on similar principles and performance characteristics (e.g., specificity, accuracy, precision). (17)
Risk Assessment
A validation risk assessment was performed by ranking identified gaps according to low, medium, or high risk. Identified gaps not covered by published data, gaps that cannot be justified, or any gap with greater severity than low were assessed during the method validation. The risk assessment identified over 40 potential risk factors related to the rFC method use. Gaps were classified in seven categories and are listed in Table 2.
No gaps were identified for most categories due to the available data and detection controls in place. Many risks were considered low due to their detectability during the method suitability and assay controls performed for each assay and incoming test materials.
The reagent hold time is an example of sample and reagent handling risk. There was no data or previous experience to support a hold time of opened reagents, resulting in a medium ranked gap. Therefore, an rFC reagent hold time study was carried out as part of the general method validation to generate data and experience with the particular reagent hold time conditions.
Roche General Product-Independent Validation
At Roche’s pilot laboratory, the objective of the validation strategy is the implementation of rFC as the primary bacterial endotoxin method while the LAL method would serve as a contingency method for the testing of water samples. This Roche site uses the water to manufacture products that are globally licensed. Therefore, the test for endotoxins must meet the requirements for multiple pharmacopoeias.
Alternative methods to those listed in USP<85> for the detection of endotoxins must be validated per USP <1223> Alternative Microbiological Methods and <1225> Validation of Compendial Procedures, in accordance to USP<1085>. The 2012 FDA Guidance for industry on Pyrogen and Endotoxins Testing allows the use of an alternative assay that provides advantages to those in the USP. This guidance further states that such alternative methods should be validated as described in USP<1225>.
General product-independent validation addresses parameters robustness, specificity, linearity, range, the limit of quantitation, accuracy, precision, and equivalence. Table 3 lists the validation parameters and the evaluation approach.
Robustness was evaluated for risk conditions that were ranked higher than low. Parameters ranked low in the risk assessment (i.e., specificity, linearity, range, the limit of quantitation, accuracy, precision, and equivalence) were not evaluated separately in the general validation.
Nevertheless, equivalency cannot be studied independently and is considered not sufficiently justified due to the lack of statistically representative data. Therefore, equivalency to LAL is evaluated by the statistical hypothesis test for non-inferiority with respect to accuracy and precision using relevant pharmaceutical water samples spiked with low, medium, and high concentrations of RSE. This demonstration of equivalency is adequate to limit the risk to patient safety and is suitable for demonstrating specificity according to USP<1225>.
The general method validation does not take into account samples containing autochthonous endotoxin (i.e., non-standardized naturally occurring) with comparability parameters. Instead, samples used are spiked with USP RSE. The comparability of the rFC and LAL methods with respect to different endotoxin sources (i.e., natural occurring endotoxins) is supported with literature reviews.
The parameters equivalency, accuracy, and precision are assessed experimentally using two data sets generated in the laboratory, one data set with water for injection (WFI) samples and one data set with purified water (PW) samples. Each data set is generated in twelve independent runs performed by two different analysts on different days, where each analytical run is associated with one sample of WFI and one sample of PW, both spiked with three RSE endotoxin levels (0.01, 0.1, 1 EU/mL). In this way, the experimental setup generates measurement variability representing within-laboratory variation under routine operating conditions. Furthermore, each endotoxin level is plated three times in duplicate on an rFC-assay plate and three times in duplicate on a LAL-assay plate. Per data set, per run and concentration, three reportable measurements for the two methods, rFC and LAL, are obtained. Overall, the two data sets contain some statistical dependence structure, which is taken into account in the statistical analysis (statistical model). No PPCs were added and freshly made assay standard curves were used for each run. Refer to Figure 1 for an illustration of the experimental design.
Product Specific Validation
The minimum requirement of a product-specific validation for rFC testing is aimed at product samples (DP, DS and IPCs). Product-specific validation is the inhibition and enhancement test or what is called method suitability. Testing is performed on three batches of each sample type using rFC and LAL. This is required and performed per USP<85>, EP 2.6.14, 5.1.10, and 2.6.32. PPCs are included with each batch of sample type tested. Beta-glucan blocking buffer is not necessary to overcome any possible interference with the rFC method as this reagent does not include Factor G that would cause false-positive results. Treatment with a dispersant is not compatible with rFC measurements due to fluorescence interference.
Statistical Analysis of Product-Independent Validation
The statistical analysis aims to demonstrate statistically significant non-inferiority, of the rFC method compared to the LAL method, with respect to accuracy and precision.
Why non-inferiority?
Equivalence, as a validation parameter, is seen in terms of assessing the safety of an article (USP<1223>). As illustrated in Figure 2, there is concern about the patient's safety in a situation of “inferiority”. On the other hand, there should be no such concern under a situation of “non-inferiority”, when rFC performs slightly worse, equal to or better than the LAL-method, since a (much) more sensitive (Figure 2(a)) or precise (Figure 2(b)) rFC performance, but also a slightly worse rFC performance, does not compromise patient’s safety.
A statistical non-inferiority hypothesis test, which decides between the two situations of “inferiority” (H0 hypothesis) and “non-inferiority” (H1 hypothesis), can limit the risk (to patient safety) of falsely deciding in “non-inferiority” to a very low probability (usually 5%). It is therefore a useful tool to verify the validation parameter “equivalence” according to USP<1223>.
Verifying equivalence with respect to accuracy implies the study of the distribution of measurement values with respect to location (i.e. the mean.). Motivated by PDA TR 33(18), the non-inferiority hypothesis H1 “rFC accuracy is non-inferior to LAL accuracy” relates to the mean recoveries of the rFC and the LAL method as follows:
The acceptance criterion for non-inferiority is derived from the hypothesis test decision rule of rejecting the hypothesis H0, that is, requiring that the lower limit of the one-sided 95% confidence interval for the ratio of the mean recoveries (blue horizontal line Figure 2(a)) is greater than 70% (orange vertical line Figure 2(a)).
Verifying equivalence with respect to (intermediate) precision implies the study of the distribution of rFC measurement values with respect to variability. The corresponding non-inferiority hypothesis H1 states that “rFC intermediate precision is better than some claimed precision”, a value which was motivated through consideration of PDA TR 33 and USP<1010>, Analytical Dara – Interpretation and Treatment. Again, the acceptance criterion for “non-inferiority” is derived from a confidence interval, namely requiring that the upper limit of the one-sided 95% confidence interval for the total rFC variability (blue horizontal line Figure 2b) is smaller than the claimed precision value (orange vertical line Figure 2b).
Conclusion
The use of statistical hypothesis testing to address the parameters equivalency, accuracy and precision makes it possible to statistically limit the risk to patient safety of falsely concluding non-inferiority of the rFC method compared to the LAL method.
A risk assessment including literature reviews helped justify the parameters specificity, linearity, range and limit of quantitation. The risk assessment also helped to identify a few easy to address gaps (i.e., medium risks) for ensuring sufficient robustness.
This validation strategy offers new rFC users a simple approach to address a full validation while minimizing the workload in the laboratory through leveraged knowledge from literature. Different from current published studies, this presented validation approach uses statistical non-inferiority hypothesis testing based on pharma water samples spiked with an endotoxin standard (as opposed to water samples containing autochthonous endotoxins).
References
- Bolden J (2019). Recombinant Factor C. In K. Williams, Endotoxin Detection and Control in Pharma, Limulus, and Mammalian Systems. (Chapter 13). Springer.
- Tindall B, Demircioglu D, Uhlig T. (2021). Recombinant bacterial endotoxin testing: a proven solution. BioTechniques, pp. Vol. 70, No. 5.
- Loverock B, Simon B, Burgenson A, Baines A. (2010). STIMULI TO THE REVISION PROCESS: A Recombinant Factor C Procedure for the Detection of Gram-negative. Pharmacopeial Forum, p. Vol. 36(1).
- Akers J, Guilfoyle D, Hussong D, McCullough K, Mello R, Singer D, et al. (2020). Functional Challenges for Alternative Bacterial Endotoxins Tests Part 2: Comparability. American Pharmaceutical Review.
- Bolden J, Knutsen C, Levin J, et al. (2020). Currently available recombinant alternatives to horseshoe crab blood lysates: Are they comparable for the detection of environmental bacterial endotoxins? A Review. PDA Journal of Pharmaceutical Science and Technology, pp. 74 (5) 602-611.
- Bolden, J, Knight M, Stockman S, Omokoko B. (2017). Results of a harmonized endotoxin recovery study protocol evaluation by 14 BioPhorum Operations Group (BPOG) member companies. Biologicals, pp. 48, 74 - 81.
- Kikuchi Y, Haishima Y, Fukui C, Murai T, Nakagarwa Y, Ebisawa A, et al. (2017). Collaborative Study on the Bacterial Endotoxins Test Using Recombinant Factor C-based Procedure for Detection of Lipopolysaccharides. Pharmaceutical and Medical Device Regulatory Science, pp. 48(4), 252–260.
- Kikuchi Y, Haishima Y, Fukui C, Nakagawa Y, Ebisawa A, Morioka T, et al. (2018). Collaborative Study on the Bacterial Endotoxins Test Using Recombinant Factor C-based Procedure for Detection of Lipopolysaccharides, part 2. Pharmaceutical and Medical Device Regulatory Science, pp. 49(10), 708–719.
- Piehler M, Roeder R, Blessing S, Reich J. (2020). Comparison of LAL and rFC Assays—Participation in a Proficiency Test Program between 2014 and 2019. Microorganisms, pp. 8(3), 418.
- Marius M, Vacher F, Bonnevay T. (2020). Comparison of bacterial endotoxin testing methods in purified pharmaceutical water matrices. Biologicals, pp. 67, 49–55.
- Muroi M, Ogura N, Mizumura H, Aketagawa J, Oda T, Tanamoto K. (2019). Application of a Recombinant Three-Factor Chromogenic Reagent, PyroSmart, for Bacterial Endotoxins Test Filed in the Pharmacopeias. Biological and Pharmaceutical Bulletin, pp. Vol. 42, No. 12.
- Bolden J, Smith K. (2017). Application of Recombinant Factor C Reagent for the Detection of Bacterial Endotoxins in Pharmaceutical Products. PDA Journal of Pharmaceutical Science and Technology, pp. Vol. 71, No. 5, 405-412
- Williams K. (2020). LAL and rFC Comparison Study Caveats. American Pharmaceutical Review.
- Abate W, Sattar A, Liu J, Conway, Jackson S. (2017). Evaluation of recombinant factor C assay for the detection of divergent lipopolysaccharide structural species and comparison with Limulus amebocyte lysate-based assays and a human monocyte activity assay. Journal of Medical Microbiology, pp. 66:888–897.
- Uhlig T, Williams K, Tindall B. (2021). rFC Validation - Simpler than you thought. PDA Journal of Pharmaceutical Science and Technology
- Eu Ph 5.1.10, Guidelines for Using the Test for Bacterial Endotoxins. 01/2021:50110
- Manual of Policies and Procedures, Center for Drug Evaluation and Research. Acceptability of Standards from Alternative Compendia (BP/EP/JP). MAPP 5310.7 Rev 1.
- Parenteral Drug Association, Inc., Evaluation, Validation and Implementation of Alternative and Rapid Microbiological Methods. Technical Report No. 33, Revised 2013
Author Biographies
Evelyn Der, Senior Scientist, Roche Genentech, QC Technology Innovation and Implementation Group, South San Francisco, CA, USA.
Carmen Marin, QC Scientist Microbiology, F. Hoffmann La Roche Ltd, Kaiseraugst, Switzerland.
Viviane Grunert da Fonseca, PhD, Non-clinical Statistician, Roche, MSAT (Manufacturing, Science and Technology), Penzberg, Germany.
Lindsey Silva, PhD, QC Director, Roche Genentech, Microbiology, South San Francisco, CA USA.
Acknowledgments
The authors would like to offer thanks to Dr Amela Wolf, Dr Boris Zimmermann, Dr Sven Deutschmann, Genentech Analytical Operations, and QC Kaiseraugst Operations for providing review and their generation of the data.
Subscribe to our e-Newsletters
Stay up to date with the latest news, articles, and events. Plus, get special offers
from American Pharmaceutical Review – all delivered right to your inbox!
Sign up now!