[Human Vaccines 1:5, 191-197, September/October 2005]; ©2005 Landes Bioscience
Correlation between Mouse Potency and In Vitro Relative Potency for Human Papillomavirus Type 16 Virus-Like Particles and Gardasil® Vaccine Samples Research Paper
ABSTRACT
CE
.D
ON
OT D
IST
RIB
UT E
.
An in vitro relative potency (IVRP) assay has been developed as an alternative to the mouse potency assay used to release Merck’s human papillomavirus (HPV) vaccine, Gardasil®, for early phase clinical trials. The mouse potency assay is a classical, in vivo assay, which requires 4–6 weeks to complete and exhibits variability on the order of 40% relative standard deviation (RSD). The IVRP assay is a sandwich-type immunoassay that is used to measure relative antigenicity of the vaccine product. The IVRP assay can be completed in three days, has a variability of approximately 10% RSD and does not require the sacrifice of live animals. Because antigen detection is achieved using H16.V5, a neutralizing monoclonal antibody, which binds to a clinically-relevant epitope, the relative antigenicity measured by the IVRP assay is believed to be a good predictor of in vivo potency. In this study, the relationship between immunogenicity, as measured by the mouse potency assay and antigenicity as measured by the IVRP assay, is demonstrated. Freshly manufactured and aged samples produced using two different manufacturing processes were tested using both methods. The results demonstrate that there is an inverse correlation between the IVRP and mouse potency assays. Additionally, clinical results indicate IVRP is predictive of human immunogenicity. Thus, antigenicity, as defined by the H16.V5 epitope, can be used as a surrogate for immunogenicity and the IVRP assay is suitable for use as the sole potency test for Gardasil samples.
IEN
INTRODUCTION
BIO
Merck Research Laboratories Departments of 1Bioprocess & Bioanalytical Research, 2Vaccine and Biologics Research and 3Vaccine Biometrics Research; West Point, Pennsylvania USA
ND
ES
*Correspondence to: Mary Retzlaff; Merck Research Laboratories Departments of Bioprocess & Bioanalytical Research; P.O. Box 4; West Point, Pennsylvania 19486 USA; Tel.: 215.652.3835; Fax: 215.652.7671; Email:
[email protected] Received 06/09/05; Accepted 08/20/05
LA
Previously published online as a Human Vaccines E-publication: http://www.landesbioscience.com/journals/vaccines/abstract.php?id=2126
05
KEY WORDS
©
20
Gardasil, human papillomavirus, in vitro relative potency, mouse potency, immunogenicity, stability, anitgenicity
ACKNOWLEDGEMENTS The authors would like to thank Dr. Neil Christensen of Penn State University for providing the H16.J4 and H16.V5 antibodies.
www.landesbioscience.com
Human papillomavirus (HPV) is considered to be the causative agent for cervical cancer.1,2 Cervical cancer kills more than 200,000 women each year worldwide, and although approximately 100 different HPV viruses have been identified, nearly 50% of all cervical cancers are associated with HPV Type 16.3 Because of the strong link between persistent HPV infection and the incidence of cervical cancer, significant effort has been directed toward developing a vaccine against HPV infection.1 Merck has developed an HPV vaccine based on a recombinantly expressed major viral capsid protein. The major component of the HPV viral capsid, L1, is an approximately 55 kDa monomeric protein that can self-assemble into empty capsid particles known as virus-like particles (VLPs).4 The VLPs are highly immunogenic5 and have structures similar to the native virion.6,7 In animal models, immunization with VLPs induced antibodymediated protection against papillomavirus infection and disease.8 In human clinical trials, immunization with Type 16 VLPs adsorbed to an aluminum-based adjuvant was well tolerated and prevented HPV 16 infection and cervical intraepithelial neoplasias associated with HPV 16.9-11 The materials used in Merck’s early clinical trials were released using a mouse potency assay, a traditional live-animal assay, which provides a measure of the immunogenicity of the sample. The mouse potency response depends on the properties of both the antigen and the adjuvant and, like all in vivo assays, is both time-consuming and labor intensive. In addition, the assay is highly variable, insensitive to minor perturbations of the antigen, and requires the sacrifice of 80 animals for each sample tested. Because of these drawbacks, an alternative potency test has been developed. The in vitro relative potency assay (IVRP) is an enzyme immunoassay, which uses a neutralizing monoclonal antibody to measure the concentration of functional epitopes in the vaccine sample. The antigenicity of the test sample is determined relative to a VLP-containing reference standard.
SC
M. Shank-Retzlaff1,* F. Wang1 T. Morley1 C. Anderson1 M. Hamm1 M. Brown2 K. Rowland1 G. Pancari2 J. Zorman2 R. Lowe2 L. Schultz2 J. Teyral3 R. Capen3 C.B. Oswald1 Y. Wang1 M. Washabaugh1 K. Jansen2 R. Sitrin1
Human Vaccines
191
Mouse Potency and In Vitro Relative Potency Correlation Study
In the study presented here, the correlation between the mouse potency and IVRP assays is demonstrated. Samples spanning a broad range of specific activities (IVRP:protein ratios) were generated and tested using both assays. The samples represent two different manufacturing processes (with and without reassembly) and various ages post-manufacture. For each sample, the mouse potency results were compared to the IVRP:protein ratio. The IVRP:protein ratios were calculated by dividing the measured IVRP by the total protein concentration. This value provides a measure of the number of functional epitopes per unit mass. Well formed VLPs will have more epitopes per particle than VLPs that are broken, aggregated, or mis-assembled. Thus, samples with higher IVRP:protein ratios are expected to be more potent in mice and in humans. The results of this study demonstrate that there is a strong inverse correlation between the natural logarithm of the IVRP results and the natural logarithm of the mouse potency ED50’s. Results from human clinical trials correlate with this observation and confirm that the non-reassembled samples are less immunogenic in humans. The data from this study support the use of VLP antigenicity as a surrogate for immunogenicity of the VLP and suggest that the IVRP assay provides a meaningful measurement of sample potency.
MATERIALS AND METHODS
Sample preparation. Unless otherwise noted, all chemicals and reagents were purchased from Sigma-Aldrich (St. Louis, Missouri). The HPV VLPs were expressed in Saccharomyces cerevisae.10,12,13 The cells were harvested, homogenized and treated with nuclease to digest DNA and RNA. The cell lysates were purified using cation exchange and hydroxyapatite chromatography. After purification, a subset of the samples used in this study was treated with dithiothreitol (DTT). DTT reduces the inter-molecular disulfides and disassembles the VLPs into subunits. After an appropriate incubation period, the DTT treated samples were dialyzed and the particles were allowed to reassemble. In the absence of reducing agent, the protein subunits spontaneously self-assemble into VLPs. For the purposes of this discussion, samples treated with DTT are designated “reassembled”. Samples not treated with DTT are designated “non-reassembled”. The disassembly and reassembly procedure used was similar but not identical to that previously reported in the literature.14 Both the reassembled and non-reassembled samples were formulated with Merck’s aluminum hydroxyphosphate sulfate adjuvant (MAA). Samples were formulated at target concentrations of 20–320 µg/mL protein, 450 µg/mL aluminum, 1.6 mg/mL histidine, 70 µg/mL sodium borate, 19 mg/mL sodium chloride and 100 µg/mL polysorbate 80. All samples were stored at 2-8˚C. Mouse potency testing. Example mouse potency data are shown in Table 1. To perform the mouse potency test, eight two-fold serial dilutions were prepared for each sample tested. The doses ranged from 0.025–3.2 µg/ 0.5 mL dose for non-reassembled samples and 0.0031–0.40 µg/0.5 mL dose for reassembled samples. The samples were diluted in MAA and sent to an outside vendor (Covance; Denver, Pennsylvania), where each dilution was injected into a group of ten female BALB/C mice. The mice were 6–8 weeks old at the time of the injection, and each mouse was given a 0.5-mL injection subcutaneously. A preimmunization serum sample was taken prior to inoculation, and a final serum sample was taken on Day 28. The resulting sera were tested in triplicate in an ELISA assay that is similar to published methods.15 HPV VLPs produced in insect cells (Commonwealth Serum Labs, Melbourne, Australia) were used as the capture antigen to prevent detection of non-specific antibodies directed against yeast impurities. Type-specific VLPs were diluted in phosphate buffered saline (PBS) and 100 µL of the diluted VLPs were added to each well of a 96-well microtiter (Nunc F96 Certified Maxisorp Immunoplates; Fisher Scientific, Pittsburgh, Pennsylvania). The plate was incubated overnight at 4˚C and washed twice in 20 mM Tris-HCl, 137 mM NaCl, 0.1% Tween 20 at pH 7.6 (TTBS). The plate was then blocked with 300
192
Table 1 Group
Example raw data for a single sample tested in the mouse potency assay and the resulting ED50 Dose a µg
Number of Subjects
Number of Positive Responders
Percent Responders
1
0.4
10
10
100%
2
0.2
10
10
100%
3
0.1
10
9
90%
4
0.05
10
8
80%
5
0.025
10
7
70%
6
0.013
10
2
20%
7
0.0063
10
1
10%
8
0.0031
10
0
0%
ED50 Determined by Probit (95% Confidence Interval) 0.022 µg (0.015, 0.033). aDose series shown is for a single reassembled sample. A dose range of 0.025–3.2 µg is used for non-reassembled samples.
µL per well of 10% Carnation Dry Milk (Nestle FoodTM) in TTBS and washed again with TTBS. Once the test plates were prepared, serum samples were diluted 1:1000 in TTBS containing 1% Carnation Dry Milk. Each serum sample was tested in triplicate by adding 100 µL of each pre and post-serum sample to each of three wells on the test plates. The samples were allowed to incubate for approximately one hour at room temperature. The plates were then washed six times in TTBS. Next, 100 µL of diluted goat-anti-mouse IgG-horse radish peroxidase conjugate was added to each well and allowed to incubate for one hour at room temperature. The plates were washed again six times and 100 µL of tetramethylbenzidine (TMB) was added. TMB (Pierce; Rockford, Illinois) is a colorimetric substrate which reacts with horseradish peroxidase. TMB was allowed to react with the horseradish peroxidase for 15 minutes at room temperature before the reaction was stopped by the addition of 100 µL of 2.0 N sulfuric acid. The extent of the reaction was determined by measuring the optical density at 450 nm. For each mouse, the ratio of the post-serum response to the preimmunization serum response was calculated. If the ratio was greater than 3.7 fold, the serum and the corresponding mouse was considered positive and sero-converted, respectively. This cutoff was established based on statistical analysis of a set of preimmunization and post-immunization sera from five separate mouse potency studies (80 mice per study). In each study, ten groups of mice were dosed with dilutions containing 0.05–0.8 µg of antigen. The cutoff value described above was selected to optimize both the sensitivity and specificity of the assay. The sensitivity was evaluated in terms of the probability of correctly judging a serum sample positive if the mouse had seroconverted. This probability is referred to as the probability of a true positive (PTP). The probability that a sample will be deemed negative if the mouse has not seroconverted or has not been inoculated is written 1-PFP where PFP is the probability of a false positive. In this analysis it is assumed that all mice in the 0.8 µg dose group seroconvert. This assumption is considered reasonable based on the observation that the average OD from 0.8 µg dose group was statistically significantly higher than the average OD associated with the preserum. The PTP for a given cutoff was, therefore taken to mean the probability of a positive response when given the vaccine at 0.8 µg. The PFP was taken to mean the probability of a positive response when given a dose of 0.05 µg? A plot of PTP versus 1-PFP was generated for all possible cutoff values for the 0.8 µg dose and the 0.05 µg dose. A reasonable cutoff should result in a small PFP (0.90) at the highest dose and a large probability that a randomly selected mouse responds positively when injected with 0.8 µg of antigen and negatively when not inoculated or inoculated at the lowest dose. The cutoff value that statisfies these criterion is 3.7 (data not shown).
Human Vaccines
2005; Vol. 1 Issue 5
Mouse Potency and In Vitro Relative Potency Correlation Study
Figure 1. Example mouse potency data for a reassembled sample. Eight dilutions of the test sample were administered to groups of ten mice and the percent responder calculated for each dose group. The results were than fit using a probit model and the theoretical dose that would result in 50% of the mice responding (serocoverting) is calculated. The same data is shown in tabular form in Table 1.
Using this cutoff, the number of mice in each group that seroconverted was determined. Percent responders for a single reassembled sample are shown in Table 1 and Figure 1. The data were analyzed by using a probit model to calculate the theoretical dose in µg that would result in 50% of the mice seroconverting(ED50). A positive control was tested in each run of the serology assay to ensure consistency in the analysis of the serum samples. The positive control consists of a lot of the H16.V5 ascites that is stored frozen. The positive control was diluted 1:100 in sample diluent and from this dilution, five two-fold dilutions are prepared for a total of six dilutions ranging from 1:100–1:3200. All six dilutions of the positive control were prepared and analyzed in duplicate in each run of the assay. For the assay to be considered valid, the geometric mean optical density (OD) for the duplicate 1:400 dilutions must fall within a preset range and the maximum OD for the 1:400 dilution divided by the minimum OD for the 1:400 dilution must be
