The Rat-Based Neurovirulence Safety Test for the Assessment of Mumps Virus Neurovirulence in Humans: An International Collaborative Study

Abstract

Because of the highly neurotropic and neurovirulent properties of wild-type mumps viruses, most national regulatory organizations require neurovirulence testing of virus seeds used in the production of mumps vaccines. Such testing has historically been performed in monkeys; however, some data suggest that testing in monkeys does not necessarily discriminate among the relative neurovirulent risks of mumps virus strains. To address this problem, a collaborative study was initiated by the National Institute for Biological Standards and Control in the United Kingdom and the Food and Drug Administration in the United States, to test a novel rat-based mumps virus neurovirulence safety test. Results indicate that the assay correctly assesses the neurovirulence potential of mumps viruses in humans and is robust and reproducible

Mumps virus is a negative-strand RNA virus in the Paramyxoviridae family that causes a communicable disease characterized by a number of acute inflammatory reactions, including aseptic meningitis. The highly neurotropic nature of mumps virus was established in a landmark study by Bang and Bang in 1943 that demonstrated evidence of invasion of the central nervous system (CNS) by the virus in ∼65% of cases [1]. Mumps virus remains a leading cause of virus-induced aseptic meningitis and encephalitis in unvaccinated populations [2, 3]

Because of the neurotropic and neurovirulent properties of wild-type mumps viruses, testing of live mumps vaccines for neurovirulence is required by most national regulatory organizations. Such testing has historically been performed in monkeys [4, 5], but results obtained from these tests have raised questions as to whether this assay can reliably discriminate neurovirulent from nonneurovirulent mumps virus strains [6–9 ]. The difficulty in evaluating the neurovirulence potential of mumps viruses by use of the monkey-based assay is suggested further by reports of causal links between mumps virus CNS infections and immunization with selected mumps virus vaccines [10–14 ]. This has resulted in the withdrawal from the market of some mumps vaccines, public resistance to vaccination, and, in some countries, complete cessation of national vaccination programs. Thus, as new strains of mumps vaccine continue to be licensed for clinical use, the development of a validated mumps virus neurovirulence safety test with more relevance to human disease risk remains an important international public health objective

To address this issue, a rat-based neurovirulence safety test was developed at the Center for Biologics Evaluation and Research at the Food and Drug Administration (FDA), and preliminary testing suggested that the model was capable of assessing the neurovirulence potential of mumps virus in humans [15]. To better evaluate the performance of the rat-based neurovirulence safety test, an interlaboratory collaborative study was initiated between the FDA and the National Institute for Biological Standards and Control (NIBSC) in the United Kingdom. For the present study, 12 mumps virus preparations of varying neurovirulence potential in humans were blinded and distributed, along with a standard operating procedure for performing the rat-based neurovirulence safety test, to each laboratory. At the conclusion of the study, the predictive value of the test for the assessment of the neurovirulence potential of these mumps viruses in humans was measured, and statistical analyses were performed to evaluate the comparability of the results between the 2 laboratories

Materials and Methods

Mumps virusesTwelve mumps virus preparations were used in the present study and are described in table 1. All were divided into 0.5-mL aliquots, kept at −80°C, and shipped on dry ice. The identity of the virus strains was blinded throughout the experimental phase of the study and was revealed at the conclusion of the study, for statistical analysis

Table 1

Mump virus strains used

Virus content determinationVirus concentrations were determined by plaque assay. Briefly, 1 aliquot of each virus was thawed and diluted in serial 10-fold increments, from 10⁻² to 10⁻⁶, by use of MEM that contained 2% fetal bovine serum (FBS; Quality Biologicals) as the diluent. Each dilution was incubated in duplicate in 12-well plates that contained 1–2-day–old confluent Vero cell monolayers. Plates were incubated at 37°C and in 5% CO₂ for 1 h. Viral inocula were removed, and wells were overlaid with a 42°C 1-mL mixture of equal parts Agar Nobel Gel 1.5% (Quality Biologicals) and 2× MEM that contained 10% FBS. Plates were incubated at 37°C and in 5% CO₂ for 5 days. Plates were stained following the addition of a second layer of agar overlay mixture containing 0.02% neutral red (Sigma). Plaques were counted the following day. The assays at the NIBSC were not performed under the agar overlay mixture; instead, they were performed with an overlay medium containing 0.8% carboxymethyl cellulose (NBS Biologicals). The cell sheet was then stained with methyl violet after 7 days of incubation. The virus titer was determined by calculation of the mean number of plaques for each pair of duplicate wells and multiplication by the inverse of the dilution factor and volume plated. Observed titers are shown in table 2. Each laboratory determined the inoculum dose on the basis of observed titers

Table 2

Virus titers determined by each laboratory

Inoculation of ratsPregnant Lewis rats were purchased from either Harlan Sprague Dawley (FDA) or Bicester (NIBSC). Two to 3 litters of 1-day-old rats were inoculated with 100 pfu of virus in a 10-μL volume of MEM by use of a 27-gauge needle. Each litter consisted of 8–10 pups. The inoculation site was in the left parietal area of the skull, ∼2 mm left of midline and midway between the bregma and lambda. On day 25 after inoculation, all rats were euthanized by CO₂ asphyxiation. Brains were removed and were fixed in 10% neutral buffered formalin for 1 week. Institutional guidelines for the care and use of laboratory animals were strictly followed

Brain processingFixed brains were cut in half in the sagittal plane along the anatomical midline. To each brain hemisphere, a second cut was made in the same plane 3–5 mm from the first cut. The resulting 3–5-mm–thick sections of brain tissue from each rat were placed in a cassette, were dehydrated by use of a graded series of ethanol baths (70%, 85%, 95%, 100%, 100%, and 100%) in which they were placed for 1 h in each bath, and were immersed in xylene for 2 h. Dehydrated tissues were then immersed in paraffin at 60°C for 2 h and were then mounted on a microtome with the side closest to anatomical midline facing the knife. A single 10-μm-thick section was obtained from each hemisphere at a depth of ∼0.5–1.0 mm from the surface. Paraffin-embedded tissue sections were placed on glass slides, warmed overnight, rehydrated, stained with hematoxylin-eosin (H-E), and dehydrated, and cover slips were attached with Permount (Fisher Scientific)

Neurovirulence assessmentAt the FDA, H-E–stained slides were placed on a scanner, and the scanned image was transferred to a computer. Image Pro Plus image analysis software (Media Cybernetics) was used to measure (in pixel units) the cross-sectional area of the entire brain (excluding the cerebellum) and the cross-sectional area of the lateral ventricle. As demonstrated in figure 1, the neurovirulence score (hydrocephalus severity) for each brain section is the quotient of the cross-sectional area of the entire brain (excluding the cerebellum) and the cross-sectional area of the lateral ventricle, expressed as a percentage. At the NIBSC, the identical parameters were measured manually by use of a grid square measuring 1 × 1 mm. The group mean neurovirulence score was the determination of the average of quotients from all brains per treatment group

Figure 1

Determination of neurovirulence score by use of a sagittal cross-section of the left hemisphere of a rat brain

Statistical analysisAfter all testing was completed, data were unblinded. The raw neurovirulence scores, being percentages, were subjected to arcsin transformation, for analysis of values and&rank order. A comparison between laboratories was done by the Tukey honestly significant difference (HSD) test, and P=.05 was considered to be significant;&rank was evaluated by use of the Wilcoxon 2-sample test. An analysis of variance (ANOVA) was then used to evaluate the ability of each laboratory to discriminate among 3 groups of virus strains: attenuated vaccine virus strains, partially attenuated vaccine virus strains, and wild-type virus strains. Members of these 3 groups are listed in table 1. In the first analysis, members of the attenuated and partially attenuated groups were market products only, and therefore, for the purposes of the analysis, were considered to be the “conservative” group. In the second analysis, members of the attenuated and partially attenuated groups included market products, as well as laboratory-passaged vaccine virus strains and clinical isolates (vaccine virus strains isolated from patients with mumps vaccine–associated meningitis after vaccination with Pluserix or Trivirex Urabe-AM9–containing vaccines) and were considered to be the “expanded” group. A nested model was used in which specimens were nested within a group

Results

A comparison of the arcsin-transformed mean neurovirulence scores determined for each virus at each laboratory is shown in figure 2. Although the scores differed between the laboratories, there were no significant interlaboratory differences in the neurovirulence&ranks assigned to the specimens (P=.471, for the Wilcoxon 2-sample test applied to the arcsin-transformed data). To assess the assay’s ability to distinguish neurovirulent virus strains from nonneurovirulent virus strains, virus strains were assigned to 1 of 3 groups: attenuated vaccine virus strains, partially attenuated vaccine virus strains, or wild-type virus strains. In the first analysis, members of the attenuated and partially attenuated groups consisted of market products only (the conservative group in table 1). Representative sagittal brain sections from rats inoculated with members of the 3 conservative groups of virus strains are shown in figure 3

Figure 2

Arcsin-transformed mean neurovirulence scores, as determined for each virus strain at each laboratory. FDA, Food and Drug Administration; NIBSC, National Institute for Biological Standards and Control

Figure 3

Representative sagittal cross-sections of brains from rats inoculated with attenuated mumps vaccine virus (Jeryl Lynn, left), partially attenuated mumps vaccine virus (Urabe-AM9, middle), and wild-type mumps virus strains (88–1961, right). Arrows enlarged lateral ventricles

An ANOVA of the neurovirulence scores represented by these 3 conservative groups demonstrated that, at both laboratories, attenuated strains could be distinguished from partially attenuated strains, and partially attenuated strains could be distinguished from wild-type strains (P<.001). Group comparisons were then performed by use of the Tukey HSD test (P=.05). For each laboratory, all 3 pairwise comparisons were significant (table 3)

Table 3

Results of 3 pairwise comparisons of conservative virus groups, by use of least-squares means differences and the Tukey honestly significant difference test

In the second analysis, members of the attenuated and partially attenuated groups consisted of market products, as well as laboratory-passaged vaccine virus strains and clinical isolates (the expanded group in table 1). When these groups were used in the analysis, the results were similar to those for the conservative group, in that the ANOVA of the neurovirulence scores represented by these groups demonstrated that, at both laboratories, attenuated strains could be distinguished from partially attenuated strains, and partially attenuated strains could be distinguished from wild-type strains (P<.001). Group comparisons performed by use of the Tukey HSD test (P=.05) also found that, for each laboratory, all 3 pairwise comparisons were significant (table 4)

Table 4

Results of 3 pairwise comparisons of expanded virus groups, by use of least-squares means differences and the Tukey honestly significant difference test

Discussion

The standard practice for assessment of the neurovirulence risk of mumps virus vaccines in humans is to test either the virus seed stock or the first few production lots of vaccine in monkeys. Data from 2 recent independent studies showed that mumps virus neurovirulence testing as currently performed in monkeys did not discern with statistical certainty the relative neurovirulence risk of many mumps virus strains in humans [6, 8]. However, preliminary test results of a novel rat-based neurovirulence safety test demonstrated correct identification of the relative neurovirulence risk of various mumps virus strains in humans [15]. These results suggested that the rat-based neurovirulence safety test might be a more predictive assay than the currently recommended monkey-based neurovirulence safety test

Results of the present study show that, although the neurovirulence scores differed by laboratory, the overall&amp;rank order of the scores was the same. Further, in both laboratories, wild-type viruses could be differentiated with statistical certainty from vaccine viruses, and attenuated vaccine viruses could be differentiated with statistical certainty from partially attenuated vaccine viruses

That the neurovirulence scores measured for each virus differed by laboratory was not unexpected. Because the scores were determined by use of a biological assay, the absolute neurovirulence scores for any given virus preparation tested in the rat-based neurovirulence safety test will differ from study site to study site, even if all procedures are executed in identical fashion using identical reagents. However, it should be noted that there were some differences in execution between the 2 laboratories—namely, use of different virus doses, rat breeding colonies, and brain area measurement methods. Of these, virus dose is likely to be the major contributor to the observed interlaboratory differences in neurovirulence scores, as is indicated by results of a study published elsewhere [15]. The use of different virus doses at the 2 laboratories was a result of virus potency being determined at each test site (see table 2). Fixed virus titers were not preassigned for common use, because product manufacturers and regulatory agencies that use this test will likely prepare material for inoculation according to their own potency determinations. However, in an effort to reduce interlaboratory differences in potency determinations, in future assays we plan to include a standard preparation of a positive control sample that would be assigned a defined infectivity titer. In this way, potency values obtained at each study site can be adjusted up or down, according to the performance of the standard in the potency test

Both brain area measurement methods used in the present study were found to yield comparable results, according to an interim analysis of blinded samples (data not shown), and were therefore not likely to affect test outcome. Notably, the standard operating procedure for the assay does not specify the method to measure area; any reasonable approach to the quantitation of the ratio of the cross-sectional area of the lateral ventricle to the cross-sectional area of the entire brain should be acceptable. Whether the use of different breeding colonies of Lewis rats played a significant role in the different neurovirulence scores obtained at the 2 laboratories is not known. Although inbred rat strains differ in their susceptibility and reaction to infection by some viruses, depending on their breeding histories [16, 17], preliminary data from testing different breeding lines of Lewis rats (from Harlan Sprague Dawley and Charles River Laboratories) with the same mumps virus stocks indicated no differences in neurovirulence scores (data not shown)

To assure better control for intraassay and interassay variability, reference viruses are needed for use in the assay. Thus, for future validation studies, distribution of a “low” virulence reference virus and a “high” virulence reference virus will be needed. An acceptable range in neurovirulence scores for these references will be established for use in the validation of assay performance. In addition, the neurovirulence of a test virus should not be determined according to its neurovirulence score per se, but, rather, according to its performance relative to that of the low virulence reference virus, by use of an appropriate confidence interval

In summary, on the basis of these initial validation studies, the rat-based neurovirulence safety test for mumps virus vaccine appears to be a robust, reproducible, and predictive test. Additional modifications are needed, such as the establishment of reference viruses and testing of other mumps vaccines—such as the Leningrad-3–based vaccine strains, which have been associated with reports of meningitis in vaccine recipients [10]. Further validation of the rat-based neurovirulence safety test through larger studies by additional independent laboratories are required to provide a basis for the adoption of this assay by regulatory authorities and vaccine manufacturers as a replacement for the monkey-based test for the assessment of the neurovirulence potential of candidate mumps vaccine virus strains in humans

Financial support: National Vaccine Program Office, US Department of Health and Human Services