New Generation of Cell Culture Assay for Smallpox Vaccine Potency

Abstract

The potency of smallpox vaccines produced in the 1970s was tested by titration onto chorioallantoic membranes of fertilized hen eggs (CAM assay). The potency specification commonly approved for these vaccines was a titer above 10⁸ pock-forming units per milliliter. We developed and validated a cell culture titration assay to have a more reliable potency test. The cell titration assay and the CAM assay were tested in parallel on 34 first-generation smallpox vaccine lots. These allowed us to demonstrate that a correlation does exist between the two titration techniques and to determine a new in-house specification for the cell titration method. This in vitro potency assay will allow us to test first-generation smallpox vaccines produced on the skin of living animals but will also give a hint of the potency specification that should be assigned for new generations of cell-derived smallpox vaccines.

The human variola virus as well as the vaccinia, cowpox, monkeypox, and camelpox viruses belong to the genus Orthopoxvirus, family Poxviridae, and are enveloped DNA viruses. An extensive serological cross-reactivity exists for all these viruses (6, 7, 8). Variola and monkeypox viruses are known to produce systemic diseases with a generalized rash in humans and primates. The case fatality rate of variola major was around 30%. The vaccinia virus usually produces a localized lesion at the site of inoculation into the skin (5, 7).

In 1796, Edward Jenner demonstrated that cowpox infection protected against the variola virus. Then, vaccination against smallpox started with a poxvirus isolated from a cow (named cowpox) followed by the vaccinia virus. Vaccination campaigns beginning in 1958 and reactivated in 1967 promoted by the World Health Organization (WHO) allowed the eradication of smallpox in 1977, and WHO recommended that vaccination cease in 1980 (1, 3, 4).

In most European countries, the strain of vaccinia virus used for vaccination was the Lister strain. In France, smallpox vaccines were stockpiled in the 1980s (9). The potency of this strategic stockpile was regularly controlled by the French medical agency by titration on chorioallantoic membranes of chicken embryos (CAM assay). This method was the official method recommended by WHO and the European Pharmacopoeia for potency testing, and the titer was to exceed 10⁸ pock-forming unit per milliliter (2, 11).

Because of the possibility that smallpox might be used for bioterrorism, it was decided in 2001 to recheck the potency of the entire stockpile by two different titration techniques. The aim of our study was to determine if a correlation exists between the CAM assay and the cell titration assay and also to determine a new in-house specification for the cell titration assay. Indeed, the cell titration assay seemed to be a very interesting assay compared to the CAM assay, because it is more reliable, more accurate, technically simpler, and less time-consuming.

MATERIALS AND METHODS

Vaccine and standard.

Thirty-four lots of freeze-dried smallpox vaccines were produced in the 1980s from the Lister strain. The titer of these vaccine lots was ≥10⁸ pock-forming units per milliliter. The standard RIV 6713/18 (Lister strain produced on calves) donated at the French public health laboratory in the 1980s by WHO and originating from the Rijks Instituut voor Volksgezondheid en Milieu was used as an internal standard in this study.

Vaccine lots and the standard preparation were reconstituted with 250 μl and 400 μl, respectively, of Iscove's modified Dulbecco's medium without fetal bovine serum.

Cells.

Vero cells were obtained from the European Directorate for the Quality of Medicines (Council of Europe). Vero cells were maintained on plastic tissue culture dishes in medium 199 containing 8% fetal bovine serum.

Titration by inoculation onto choriallantoic membrane of fertilized hen eggs.

Two adequate dilutions (usually 10⁻⁶ and 10⁻⁷) of each vaccine lot were run in each assay. For each dilution, 10 chicken embryos (12 days old) were inoculated with 100 μl of the dilution onto the choriallantoic membrane. After 2 days of incubation at 37°C, membranes were collected from the eggs, and the number of pocks were counted. The dilution chosen for calculation of the virus concentration (expressed in pock-forming units per milliliter) was the dilution which induced a countable number of lesions exceeding 10 per membrane.

Cell titration.

The cells were plated on 96-well plates (7,000 cells/well) in Iscove's modified Dulbecco's medium with 5% fetal bovine serum the day before titration. Three independent dilution series of the vaccine samples and of the internal standard were prepared in Iscove's modified Dulbecco's medium without fetal bovine serum. Cells were inoculated with 100 μl of a 10⁻⁵ to 10⁻⁹ dilution, and for each dilution, eight wells were infected. The 96-well plates were incubated at 37°C, and cytopathic effect was read 9 to 10 days after inoculation.

Precision of cell titration method.

The precision of the cell culture method was assessed with the reference vaccine. In order to obtain a normal distribution of the data, a logarithmic transformation was applied to the titers. The experimental design consisted of six groups of assays complying with conditions of intermediate precision (assays performed independently with the same method in the same laboratory on different days). Within each group, four assays were carried out under conditions ensuring repeatability (assays performed independently with the same method, in the same laboratory, with the same equipment, on the same day). The homogeneity of within-test variances was verified by a Cochran's test. To evaluate the repeatability and intermediate precision variances, a variance components analysis was performed.

Normality and consistency of cell titration assay.

A test was run to determine whether the distribution could be adequately modeled by a log-normal distribution. The chi-square test divides the range of the distribution into 20 equally probable classes and compares the number of observations in each class to the number expected.

A control chart was set up to analyze the consistency of the first-generation vaccine stockpile. In order to obtain a normal distribution of the data, a logarithmic transformation was applied to the titers. The control limits were calculated as m₀ ± 3s₀, where m₀ is the mean and s₀ is the estimation of the true standard deviation of the titer distribution.

Correlation between in vivo (CAM assay) and in vitro (cell titration) methods.

The relationship between each pair of variables was performed by fitting a logarithmic model: log₁₀(in vivo titer) = a + b × log₁₀(in vitro titer). This statistical study required a regression analysis, including an analysis of variance. The Pearson product moment correlation between each pair of variables was calculated. This correlation coefficient ranges between −1 and +1 and measures the strength of the linear relationship between the variables. The P value tests the statistical significance of the estimated correlations. A P value below 0.05 indicates statistically significant nonzero correlation at the 95% confidence level.

RESULTS

Precision of cell titration method.

An evaluation of the precision of the cell titration method was performed with the reference vaccine. Cochran's test showed that the variances of the six groups (assays) were homogeneous (P = 0.392). The precision results (Table 1) showed acceptable repeatability (within-test variance) and intermediate precision (between-test variance). The precision of the method was calculated as ±0.27 log₁₀.

TABLE 1.

Precision of the cell titration method, assessed by using the reference vaccine: within-test and between-test variance component analysis

Source	Sum of squares	Degrees of freedom	Mean square	Variance component	Percent
Between-test	0.0860	5	0.0172	0.0002	1.34
Within-test	0.2936	18	0.0163	0.0163	98.66
Total (corrected)	0.3796	23

Normality and consistency of cell titration assay.

Because the P value for the chi-square test performed equaled 0.06, we could not reject the assumption that the distribution (logarithmic transformation of the data) came from a normal distribution.

A control chart of vaccine production (34 results) was set up to analyze the consistency of the cell titration method (Fig. 1). The mean titer was around 8.02 log₁₀ tissue culture infectious doses (TCID₅₀)/ml. The upper and lower control limits were 9.47 log₁₀ TCID₅₀/ml and 6.57 log₁₀ TCID₅₀/ml, respectively.

FIG. 1.

Control chart of 34 smallpox vaccine lots for the cell culture method, expressed as log₁₀(TCID₅₀/ml).

Correlation between in vivo (CAM assay) and in vitro (cell titration) methods.

A statistical description of the relationship between each pair of variables was performed (Table 2 and Fig. 2).

TABLE 2.

Correlations between in vivo (CAM assay) and in vitro (cell culture titration) methods

Method	Parameter	Estimate	Source	Sum of squares	Degrees of freedom	Mean square	F ratio	P
Regression descriptiona	Intercept	5.82						0.000
	Slope	0.358						0.001
Analysis of variance			Model	0.919	1	0.9188	14.00	0.0007
			Residual	2.100	32	0.0661
			Total (corrected)	3.019	33

^aCorrelation coefficient = 0.552.

FIG. 2.

Correlation between in vivo (CAM assay) and in vitro (cell culture titration) methods. The inner bounds show 95.0% confidence limits for the regression line. The outer bounds show 95.0% prediction limits for future observations. The arrow represents the in vitro titer corresponding to a predicted in vivo titer of 8.0 log₁₀.

Since the P value in the analysis of variance table was less than 0.01, there was a statistically significant relationship between the in vivo and in vitro results at the 99% or higher confidence level. The R² statistic indicates that the model as fitted explains 30% of the variability in the in vivo results. The correlation coefficient equals 0.55, indicating a moderately strong relationship between the in vivo (CAM assay) and in vitro (cell titration) results.

Moreover, the regression analysis indicates that a titer of greater than 7.6 log₁₀ TCID₅₀/ml with the in vitro method would have been found above the in vivo specification of 8.0 log₁₀ pock-forming units per milliliter at the 95% confidence level.

DISCUSSION

The idea to propose another technique to control the potency of smallpox vaccines is not new and was developed in the 1970s (10). The eradication of smallpox and the discontinuance of vaccination in 1980 were probably the reasons why this method was no longer pursued. Because of the new possibility that the variola virus may be used as a biological weapon and that some countries have a smallpox vaccine stockpile, it is again opportune to develop and standardize a method of titration on cell culture for smallpox vaccines. Because the reference method of titration in all regulatory documents was the CAM assay (titration in chorioallantoic membranes of chicken embryos), we decided to compare the cell titration assay to the reference method.

If we consider the results, as expected, a significant relationship was observed between the in vivo (CAM assay) and the in vitro (cell titration) methods. The correlation coefficient of r = 0.552 is not particularly high (R² = 30%), as can be seen from Fig. 2. This is not surprising, given that all the vaccine lots tested were of similar potency. A large component of the variation is therefore the independent assay-to-assay variation for each method, which is not expected to be correlated. A more realistic assessment of the agreement between methods could only be obtained with data from vaccines of a wider range of potencies, including subpotent vaccines. Moreover, as shown with the control chart, the lack of trends, the low precisions, and the fact that no point is out of the control limits allowed us to consider that our in vitro control method is suitable to monitor the consistency of smallpox vaccine production.

Indeed, the consistency and the correlation studies led to define a lower in-house specification for our in vitro method, 7.6 log₁₀ TCID₅₀/ml.

Concerning the cell titration method, as expected, because of the uniformity of the experimental conditions, the precision was acceptable and usual for this type of technique (±0.27 log₁₀). The control chart implemented with our internal reference confirmed this (±0.31 log₁₀) (data not shown). The control chart became a criterion which must be verified to validate or invalidate the assay in routine testing.

Both potency methods used to assay the French stockpile of freeze-dried smallpox vaccines produced in the 1970s confirmed the very good stability of this vaccine stored at −20°C. This information is very important in view of the production of a new generation of cell-derived vaccines that could be stable as freeze-dried lots for at least 25 years for an improbable act of bioterrorism.

This new validated method of cell titration will allow regulatory agencies and also manufacturers to link the potency of first-generation smallpox vaccines to that of second-generation vaccines. The potency specification of the cell-derived vaccines could not be a transposition of the potency specification of the CAM assay but will depend on the potency control method and on its correlation with the reference method (CAM assay).

Potency assay by cell titration could be included in all international recommendations such as the European guidelines, European Pharmacopoeia monographs, and WHO recommendations. This reliable and simple method should allow performance of the potency assay throughout development and during the process of quality control of smallpox vaccines.