Preparation and qualification of internal rabies reference standards for use in the rabies rapid fluorescent focus inhibition test

The World Health Organization (WHO) international standard rabies immune globulins (SRIGs) allow the standardisation of the cell-based rapid fluorescent-focus inhibition test (RFFIT) for rabies virus neutralising antibody measurement. SRIG stocks have been depleted. We describe the preparation and qualification of two internal rabies reference standards (IRRSs), calibrated against WHO SRIGs. Candidate IRRSs IMORAB2, from human rabies immunoglobulin; and GCIRAB1, from pooled serum samples from healthy adults immunised with licensed rabies vaccine, were generated. IRRSs were qualified for use in RFFIT based on pre-determined acceptance criteria. Unitage (IU/mL) was assigned using WHO-1 and WHO-2 SRIGs as calibrators. Geometric mean concentrations (GMCs) (% geometric coefficient of variation), calibrated against WHO-1 and WHO-2 SRIGs, were: 1.8 IU/mL (18.7%) and 1.5 IU/mL (17.8%) for IMORAB2; and 2.9 IU/mL (17.5%) and 2.5 IU/mL (16.7%), respectively, for GCIRAB1. We demonstrated IRRS specificity in competition studies using homologous (inactivated Pitman Moore rabies virus) and heterologous (inactivated vesicular stomatitis virus) antigens and acceptable accuracy/linearity of WHO SRIGs using IRRSs as calibrators. Concordance between IRRS and the WHO-1 SRIG was demonstrated using (non-)clinical human serum samples. The candidate reference standards are suitable for use as IRRS in the in-house rabies RFFIT. Funding:Sanofi Pasteur.

The WHO-1 SRIG R-3 (59 IU/mL) and WHO-2 SRIG RAI (30 IU/mL), established in 1984 and 1994, respectively, have been used at the Sanofi Pasteur (SP) Global Clinical Immunology (GCI; Swiftwater, USA) department for measuring RVNA levels in clinical trial samples. The WHO-1 SRIG, the first rabies immunoglobulin reference serum of human origin, was prepared from pooled sera from vaccinated humans 5 . A second pool of sera from vaccinated humans was tested in 1993 (WHO-2) 6 . However, as stocks of the WHO-1 and WHO-2 SRIGs are now depleted, replacement rabies reference standard preparations are needed for testing. In this study, two potential internal rabies reference standards (IRRSs), IMORAB2 and GCIRAB1, were created and were calibrated against the WHO-1 and WHO-2 SRIGs for use in the SP in-house RFFIT. These new reference standard preparations ensure continued direct comparability of RVNA level measurements over time within the laboratory.

Results
Assignment of unit values. The observed geometric mean concentrations (GMCs) for IMORAB2 lot #1 were 1.8 IU/mL using WHO-1 SRIG as the calibrator and 1.5 IU/mL using WHO-2 SRIG as the calibrator (Table 1; Fig. 1a). For GCIRAB1, unit values were first assigned to the undiluted pooled preparation (Supplementary Table S1). Based on the assignment value of the undiluted material, GCIRAB1 was diluted to approximately 2.0 IU/mL. The observed GMCs for diluted GCIRAB1 lot #1 were 2.9 IU/mL using WHO-1 SRIG as the calibrator and 2.5 IU/mL using WHO-2 SRIG as the calibrator (Table 1; Fig. 1b). The unitage in IU/mL assigned using the WHO-1 SRIG was used for the qualification of the candidate IRRSs.

IRRS Candidate Calibrator
Concordance. The 90% CI of the concordance slope was within the acceptable interval of 0.80-1.25 for clinical and non-clinical samples, and clinical and non-clinical samples combined, for both candidate IRRSs (Fig. 5). The 90% CI of the percent differences in the RFFIT GMC values between those generated using the WHO-1 SRIG and those generated with IMORAB2 or GCIRAB1 as calibrator were within the acceptable range, of −30% to 30% (  Table S5).

Discussion
The use of a reference standard in laboratory assays is essential for the calibration and harmonization of assay data, allowing for meaningful interpretation of results and comparisons between laboratories and between studies over time [7][8][9][10] . The RFFIT assay is complex due to the fact that sensitivity and specificity are affected by multiple factors, lending to potential variability between results obtained from different laboratories. The harmonization of the RFFIT is particularly important as the originally described method 11 has been modified such that different assay formats are currently used [12][13][14] .
In this study, candidate IRRSs, IMORAB2 and GCIRAB1, were prepared and calibrated using the WHO-1 and WHO-2 SRIGs, for use in the SP in-house rabies RFFIT method. Pre-defined acceptance criteria were met for the qualification of both candidate IRRSs, which included: assessment of the precision of the assigned unitage and established ED 50 titre ranges; specificity of the IRRSs in competition with homologous and heterologous antigens; the accuracy and dilutional linearity of the WHO SRIGs when using the candidate IRRSs as calibrators; and evaluation of the concordance of results obtained with the IRRSs with those obtained using the international WHO-1 SRIG. Indeed, results using IMORAB2 and GCIRAB1, with either clinical or non-clinical samples, showed good concordance with historical data generated during a previous rabies vaccine clinical trial or through prior sample testing using WHO-1 as a calibrator.
The reference standard should reflect how a test sample would behave in the assay, across a range of assay formats 10 . Pooled serum samples from several individuals who have been exposed to the pathogen, via infection or vaccination, provide a widened spectrum of antibody reactivates and more closely represent a test sample matrix. As such, they are more likely to work in a range of assay formats and are thus considered to be an ideal basis for an antibody assay standard 10 . While our results showed that the GCIRAB1, generated from pooled sera, is suitable for use as an internal reference standard for RVNA testing with the RFFIT, we also showed that the candidate IRRS prepared from commercially available human rabies immunoglobulin would be suitable for use as well. The use of purified rabies immunoglobulin rather than serum samples would reduce the levels of biosafety containment required for handling samples in the laboratory. Although it is feasible to obtain rabies immunoglobulin, it's matrix would in theory be more limited than that of pooled sera 10 .
The assigned values for the potency of the WHO-1 and WHO-2 SRIGs (59 and 30 IU/mL, respectively) are important in providing known values against which we were able to calibrate the candidate IRRSs for use in standardized in-house testing. As they are not exactly equivalent and have both been used extensively in different studies, both WHO SRIGs were used for calibration of the candidate IRRSs to allow for comparability.

Parameter tested Acceptance criteria
Assignment of unit value (IU/mL) The precision should not exceed a GCV of 30%.

Establishment of ED 50 titre range
The assigned range is the 10 (mean ± 2 SD) where mean and standard deviation (SD) are calculated using log 10 -transformed ED 50 titres.

Competition with homologous antigen
The highest concentration of the homologous competitor must decrease the rabies virus neutralising antibody (RVNA) titre values by ≥80%.

Competition with heterologous antigen
Heterologous (unrelated) competitor must not cause the RVNA titre to reduce by>30% as compared to the RVNA titre obtained without any competition (baseline control).
Accuracy/linearity 80% of the spiked WHO-1 and WHO-2 SRIGs with results ≥ LLOQ, must have percent recovery of 70-130% for both WHO-1 and WHO-2 SRIGs. Linear regression slope (observed GMC vs. expected) must be 0.80-1.25 and the coefficient of determination (R 2 ) must be ≥ 0.95 for both WHO-1 and WHO-2 SRIGs.

LLOQ
The calculated %GCV for the positive samples near LLOQ must be ≤30%. All negative serum must remain negative.

Concordance
The 90% confidence interval (CI) of the concordance slope must be within the range of 0.80-1.25. The 90% CI of the percent difference must be within the range (−30%, 30%). www.nature.com/scientificreports www.nature.com/scientificreports/ The importance of the WHO recommendation to use a single recognized reference standard for assay validation 13 has been shown in a previous study comparing RFFIT performance with different standard sera. Titres obtained using standards prepared by the National Institute for Food and Drug Control of China were www.nature.com/scientificreports www.nature.com/scientificreports/ significantly different to titres obtained using WHO international standards 15 . It should also be noted that the potency assigned to a reference standard by one method, e.g. the RFFIT, may not be the same in a different method such as an ELISA method 16 . The unit values assigned for the candidate reference standards described will thus be applicable specifically to the RFFIT method in which it was qualified.
The shelf-life and stability of the reference standard are important aspects to be considered and ideally optimized such that assay performance can be monitored with the same standard over a number of years 17 . However, the degradation of prepared references can occur over time despite being stored under controlled conditions. The stability of the internal references evaluated in this study will therefore need to be monitored by RFFIT testing, with results compared to pre-established baseline results, using in-run controls and performance monitoring of sample panels to monitor any drift over time. Regular monitoring of assay performance using the reference standard and pre-defined criteria is also essential to avoid drift in assay performance, and to control for potential modifications over time 10,13 .

Conclusion
In the current study, we demonstrated the evaluation and qualification of two IRRSs, prepared using two different approaches. Both the candidate IRRSs IMORAB2, prepared from human rabies immunoglobulin, and GCIRAB1, prepared from pooled RVNA positive human serum samples, met pre-determined acceptance criteria. These reagents are both suitable for use as internal reference standards in the SP in-house rabies RFFIT method; their use will ensure continued direct comparability of RVNA level measurements over time, and between studies, within the laboratory.
Methods preparation of iRRSs. Two different approaches were used for the preparation of the IRRSs. IMORAB2 was prepared from human rabies immunoglobulin produced by SP (Imogam, 187 IU/mL) that was diluted with serum-free cell culture medium to a target GMC of approximately 2.0 IU/mL (a total volume of approximately 140 mL was prepared). The diluted candidate IRRS was assigned a lot number (e.g., lot #1) and aliquoted at 240 μL/vial for storage at −40 °C to −80 °C; the stability and performance of the diluted material was monitored in each RFFIT assay run.
GCIRAB1 was prepared from pooled human serum samples positive for RVNA. Samples from nine donors vaccinated with the licensed human diploid cell rabies vaccine (Imovax Rabies) several years before sample donation were screened using the in-house RFFIT; four donors were selected based on the level of RVNA in the sample. Selected samples were pooled and aliquoted (undiluted) at 5 mL/tube and stored at −40 °C to −80 °C (total volume, 250 mL). 24 mL of GCIRAB1 (undiluted) was diluted with serum-free cell culture medium to a target GMC of approximately 2.0 IU/mL. The diluted candidate IRRS was assigned a lot number (e.g., lot #1) and aliquoted at 240 μL/vial for storage at −40 °C to −80 °C; the stability and performance of the diluted material was monitored in each RFFIT assay run.
All experiments and experimental protocols conducted to evaluate and qualify the IRRS were approved by SP, Swiftwater, PA.
Rffit method. The RFFIT, initially described by Smith et al. (1973) 11 , was implemented under optimised and validated conditions 18 . The RVNA titre (50% neutralisation, ED 50 ) of a test sample was mathematically interpolated using the Reed and Muench method 19 and was calibrated and converted into IU/mL against the WHO SRIG or the candidate IRRS. cells and rabies virus. Baby hamster kidney (BHK)-21 cell banks (American Type Culture Collection, catalog #CL-10) were qualified before use in in-house RFFIT and cultured as described previously 18 . Challenge virus standard 11 (CVS-11), lot #MLE 04-2015, was produced by SP at Marcy L'Etoile using virus stock received from Centers for Disease Control and Prevention, Atlanta. The RABV CVS-11 strain was diluted to target a challenge virus dose of 50 (tissue culture infectious dose [TCID] 50 )/100 μL and back-titrated to determine the actual TCID 50 for each assay run and was qualified for use in in-house RFFIT as described previously 18,19 .
WHo SRiGs. The WHO-1 SRIG (lot R-3; 59 IU/mL) was obtained from the Food and Drug Administration/ Center for Biologics Evaluation and Research; WHO-2 SRIG (lot RAI; 30 IU/mL) was obtained from the National Institute for Biological Standards and Controls. Both were diluted to 2.0 IU/mL according to manufacturer's instructions, and heat inactivated at 56 °C for 30 min prior to use. test serum samples. Human serum samples were obtained from healthy adult SP employees: RVNA-positive samples were obtained from those immunised with human diploid cell vaccine (Imovax Rabies); RVNA-negative samples were obtained from unvaccinated individuals. Samples were prepared, coded, and randomised (blinded) by a biostatistician. Procedures complied with Health Insurance Portability and Accountability regulations and Sanofi Policies and Procedures. Informed consent was obtained from these volunteers to collect samples for research use. Twelve 150 μL single-use aliquots were created for each sample, and heat inactivated at 56 °C for 30 min before use.
An additional subset of clinical serum samples from individuals who had participated in a previous SP purified Vero rabies vaccine clinical study (NCT03145766), was used for concordance studies. The study was carried out in accordance with the Declaration of Helsinki and the International Conference on Harmonisation guidelines for Good Clinical Practice. Written informed consent to use clinical samples for future research purposes was obtained during the clinical trial following the study protocol approval by the local Institutional Review Board (IRB). Only those samples for which participants had provided informed consent for future research use were included in the current study. Sixty samples with RVNA titres ranging from <0.2 IU/mL to 8.0 IU/mL were