- Email: [email protected]
Points to consider in the development of a surrogate for efficacy of novel Japanese encephalitis virus vaccines
Vaccine 18 (2000) 26±32
www.elsevier.com/locate/vaccine
Discussion
Points to consider in the development of a surrogate for ecacy of novel Japanese encephalitis virus vaccines Lewis Marko 1 Laboratory of Vector-borne Virus Diseases, Center for Biologics Evaluation and Research, Food and Drug Administration, Rockville, MD20852-1448, USA
Abstract Although an eective killed virus vaccine to prevent illness due to Japanese encephalitis virus (JEV) infection exists, many authorities recognize that a safe, eective live JEV vaccine is desirable in order to reduce the cost and the number of doses of vaccine required per immunization. A large-scale clinical ecacy trail for such a vaccine would be both unethical and impractical. Therefore, a surrogate for the ecacy of JE vaccines should be established. Detection of virus-neutralizing antibodies in sera of vaccinees could constitute such a surrogate for eacy. Field studies of vaccinees in endemic areas and studies done in mice already exist to support this concept. Also, titers of virus-neutralizing antibodies are already accepted as a surrogate for the ecacy of yellow fever virus vaccines and for the ecacy of other viral vaccines as well. In developing a correlation between N antibody titers and protection from JEV infection, standard procedures must be validated and adopted for both measuring N antibodies and for testing in animals. A novel live virus vaccine could be tested in the mouse and/or the monkey model of JEV infection to establish a correlation between virus-neutralizing antibodies elicited by the vaccines and protection from encephalitis. In addition, sera of subjects receiving the novel live JEV vaccine in early clinical trials could be passively transferred to mice or monkeys in order to establish the protective immunogenicity of the vaccine in humans. A monkey model for JEV infection was recently established by scientists at WRAIR in the US. From this group, pools of JEV of known infectivity for Rhesus macaques may be obtained for testing of immunity elicited by live JE vaccine virus. Published by Elsevier Science Ltd.
1. Introduction The ecacy of a vaccine is de®ned as the relative ability of the vaccine to induce a protective immune response. Ecacy is classically measured by the results of a clinical trial in which a reduction of disease incidence in an at-risk population is directly proven. Most clinical trials designed to demonstrate vaccine ecacy are doubly blinded and placebo-controlled (DBPC). This is an essential element in the plan of such a trial, if the intent is to produce direct evidence for ecacy. A DBPC study of the ecacy of formalin-inactivated 1 The opinions expressed herein are solely those of the author and do not necessarily represent the ocial policy of the US Food and Drug Administration
0264-410X/00/$ - see front matter Published by Elsevier Science Ltd. PII: S 0 2 6 4 - 4 1 0 X ( 0 0 ) 0 0 0 3 8 - 4
Japanese encephalitis virus (JEV) vaccines was conducted in Thailand, in 1984±85 [1]. About 65,000 schoolchildren, ages 1±14, were randomly divided into three groups and immunized with two doses of vaccine derived from the Nakayama strain of JEV, or vaccine derived from a mixture of Nakayama and Beijing strains of JEV, or placebo. The Beijing strain of JEV most closely resembles strains of JE that currently circulate in Thailand and the remainder of southeast Asia, while the Nakayama strain was isolated in the 1930s and has been propagated in the laboratory since then. The children were monitored for vaccine safety and illness. The diagnosis of JEV-related illness was made by the detection of JEV-speci®c IgM in either serum or cerebro-spinal ¯uid (CSF) in children with fever. During the post-immunization surveillance period, there were 51 cases of JE per 100,000 popu-
L. Marko / Vaccine 18 (2000) 26±32
lation in the placebo group vs ®ve cases of JE per 100,000 in the vaccinated group. This gave a calculated ecacy of 91% with a 70±97% con®dence interval (CI) for both vaccinated groups vs the group that received placebo. The study also demonstrated that any antigenic dierences between the Nakayama and Beijing strains of virus used to prepare the vaccines were not consequential for ecacy; children immunized with the Nakayama-strain-only vaccine were as well protected as children who received the bivalent formulation. The results of this trial were a pivotal element in the approval process for the Nakayamastrain JE vaccine that is licensed in the US. There is no absolute standard for the level of ecacy of a vaccine that is a candidate for licensure in the US. Historically, only the ecacy of poliovirus vaccines has ever been mandated by law. The Code of Federal Regulations (CFR) stipulated that such vaccines must protect 90% of vaccinees with 95% con®dence, but this section of the CFR was deleted in 1996. A variety of dierent vaccines with a wide range of demonstrated ecacy have recently been licensed in the US. The acellular pertussis vaccines, Certiva, Tripedia, and Infanrix were shown to be 71%, 80%, and 83% ecacious, respectively. A new rotavirus vaccine, Rotashield, was shown to be 90% ecacious in the prevention of severe rotavirus gastroenteritis but only 50% ecacious in the prevention of mild signs of rotavirus infection. Two inactivated hepatitis A virus (HAV) vaccines, Havrix and Vaqta, were at or near 100% ecacy in the prevention of illness. Factors that are considered in determining whether the level of ecacy of a new vaccine is acceptable for licensure include the availability of an alternative vaccine, the ecacy and other qualities of that alternative vaccine (e.g. cost, adverse event pro®le, stability), and the prevalence and severity of the disease that is to be prevented by the new vaccine. 2. Surrogates for vaccine ecacy A classical ecacy trial of a novel vaccine may be impractical in the modern era, for all or some of the following reasons: (i) vaccines to prevent many of the major infectious diseases are already available. Often the study vaccine will not be the only vaccine with the potential to prevent disease caused by the targeted pathogen. If the alternative vaccines are already licensed, it may not be ethical to perform an ecacy trial that includes a placebo control group. This consideration currently pertains to the development and testing of new JEV vaccines. (ii) Due to a pre-existing vaccine eect, the incidence of disease may be so low that a prohibitively large study would be required to adduce a statistically signi®cant result. For example,
27
the incidence of JE clinical illness in Southeast Asia is currently estimated to be 10 cases per 100,000. Given this incidence of disease, it would require 247,000 subjects per group (vaccinees vs placebo) to demonstrate 90% ecacy with a CI similar to that which was demonstrated in the Thai study (Lachenbruch, personal communication). Such a clinical trial would be prohibitively costly and dicult to perform. (iii) It may be impossible to design a valid DBPC clinical trial to establish the ecacy of a combination vaccine in a population that may have widely variant levels of pre-existing immunity to one vs another of the targeted pathogens or in which the inclusion of a placebo group is not acceptable for the same reasons applicable to the study of monovalent vaccines. In order to circumvent the problems associated with executing a DBPC clinical trial for vaccine ecacy, regulatory authorities may permit the substitution of data establishing a ``surrogate'' for ecacy. This is a laboratory test result, usually a serologic assay, that has an accepted positive correlation with the induction by a vaccine of a protective immune response in vaccinees. The induction of a certain minimum titer of antibodies in vaccinees, typically as detected by a speci®c RIA or ELISA, a virus-neutralization (N) assay, or an assay for hemagglutination-inhibiting (HI) antibodies, is often taken as a surrogate for a protective eect of a vaccine. For example, the N assay provides a surrogate for ecacy of poliovirus vaccines; a titer of >1:8 is taken as indicative of protection against poliovirus infection. ELISAs or RIAs using HAV as antigen provide surrogates for ecacy of hepatitis A and B vaccines. In contrast, no serologic assay has been shown to correlate with the ecacy of rotavirus or pertussis vaccines. Where a surrogate for ecacy has been identi®ed, it may be sucient to demonstrate that a new vaccine stimulates a protective immune response in a large and heterogeneous cohort of seronegative subjects. One major advantage of relying upon a surrogate to show protection, as opposed to a classical ecacy trial, is that a study using a surrogate may be carried out in a population that is not at risk for the targeted disease. This simpli®es study design and reduces safety concerns. If a similar licensed vaccine already exists, one form that such a study may take is that of a random, doubly-blinded ``head-to-head'' comparison between the trial vaccine and the previously licensed product. Issues of safety as well as ecacy can be addressed in such a trial. A surrogate for ecacy may be supported by careful analysis of the results of a previously conducted clinical trial designed to demonstrate ecacy directly. For example, a controlled DBPC study was conducted in 1083 hepatitis B (HB) seronegative homosexual males to show ecacy for Heptavax, the ®rst HB vaccine to
28
L. Marko / Vaccine 18 (2000) 26±32
be approved in the US [2]. About half the subjects were immunized, and all were monitored for 24 months for the occurrence of subsequent HB infection. The overall incidence of the disease was 25.6% in the placebo group and 3.2% in the vaccinated group. Further analysis of the cases among vaccinees revealed a negative correlation between antibody titer after immunization and disease occurrence. Among 21 vaccinees who had titers R2.1 RU (Ausab Ratio Units), there were seven cases, for an incidence of 33.3%. This was essentially not dierent from the incidence in the placebo group. Among nine subjects who had titers between 2.1 RU and 10 RU, there was one case, for an incidence of 11.11%. And among 421 vaccinees who had titers r10 RU, there was also only one case, for an incidence of 0.23% and nearly 100% vaccine ecacy in this group. This result established the concept that the titer of antibody (r10 RU or mIU/ml) was a correlate of protection. Heptavax was a ``®rst-generation'' vaccine; it was composed of HBsAg particles puri®ed from the sera of infected donors. Heptavax was superceded in the US by vaccines derived via recombinant DNA technology in yeast. The immunizing antigen in these second-generation HB vaccines (Recombivax and Engerix B) was shown by sponsors to be physically identical to the antigen in Heptavax by available methodologies. However, when the ®rst of these vaccines (Recombivax) was under review, a small-scale additional ecacy study (in infants borne of mothers who were chronic carriers of HB [3]) was required as a supplement to an immunogenicity study demonstrating that Recombivax induced antibody titers r10 mIU/ml in an Ausab-like assay in nearly 100% of 1500 adult subjects who received the recommended vaccine dosage. Thus, a surrogate for ecacy based on data obtained with a vaccine derived using one speci®c technology will not automatically be acceptable for a vaccine derived using an entirely dierent one, though the target disease and even the antigen content of the vaccine may be the same in both cases. After the Recombivax approval process, FDA required no formal ecacy trial in support of approval of the second yeast-derived HB vaccine, Engerix B; the immune correlate of protection was accepted. Animal studies can be used to provide support for a surrogate for ecacy of a vaccine, especially where there exists an animal model that mimics pathogenesis of the disease in humans. Evidence of this kind was available in support of the use of RIA or ELISA results as a surrogate for ecacy of hepatitis A (HA) vaccines. Chimps were used as a model for the disease in humans. It was initially shown that chimps passively immunized with pooled hyper-immune serum from HA-infected humans were protected from challenge. Next, chimps were challenged at dierent time inter-
vals after passive immunization. As antibody titers waned, these chimps remained protected as long as they had serum antibody levels that were at or above detectability using a speci®c RIA (HAVAB or modi®ed HAVAB, available in kit form from Abbott Labs) or by a speci®c ELISA, ``ELISA-Ig'' (Smithkline-Beecham Biologicals) [4]. Both are competition assays with identical cut-os, using puri®ed virions as antigen. In additional studies, it was shown that the antibodies in human sera detected in the ELISA-Ig could be completely blocked by a mixture of two monoclonal antibodies that recognized non-overlapping epitopes with virus-neutralizing activity [5]. This established a correlation between ELISA antibodies and N antibodies elicited by HA infection. The correlation was borne out by studies of sera from subjects who had naturally acquired HA infection or who had been immunized with the inactivated HA vaccine, Havrix. In summary, animal and laboratory studies established that ELISAmeasured antibodies had N activity and that vaccineinduced antibodies were qualitatively similar to those resulting from natural infection. Finally, a large-scale clinical trial to show ecacy of Havrix was conducted among more than 40,000 schoolchildren in Thailand. The results of this study were accepted as evidence for the ecacy of Havrix in support of its approval in the US. The overall outcome of the approval process for Havrix (after FDA review of the formal ecacy trial and the earlier chimp and serologic studies) supported the proposition that antibody detected in the ELISA-Ig [and more recently, the ``Enzymun'' assay (Boehringer-Mannheim]) indicates the presence of protective immunity in a vaccinee who has received an inactivated whole virus HA vaccine. This principle can now be applied in the approval process for similar HA vaccines.
3. Surrogates for ecacy of JE vaccines The past history suggests a plausible approach to establishing a surrogate for the ecacy of new JE vaccines. Elements of this process could include: (i) the selection of one or more likely immunologic surrogates for protection. (ii) The selection of an animal model for JE pathogenesis. (iii) Performing passive protection experiments in the selected animal model, when possible using human antibodies from naturally-infected subjects and from vaccinees. (iv) Correlation of the results of such studies with antibody levels as determined by the projected surrogate assay. (v) Conduct of a ``head-to-head'' immunogenicity (and safety) trial, using an existing JE vaccine, for example the USlicensed inactivated JEV vaccine (Biken/Connaught Labs), as a comparator.
L. Marko / Vaccine 18 (2000) 26±32
3.1. Evidence for one serotype of JEV Flavivirus infections elicit neutralizing and non-neutralizing antibodies that may have serotype, species, subgroup or group speci®city. Flavivirus antibodies have been commonly measured by conventional ELISAs, HI assays, assays for complement-®xing (CF) antibody, and the N assay. As normally performed, ELISA and HI and CF assays are highly cross-reactive, whereas the N assay provides the basis for speciation of ¯aviviruses [6]. Laboratory studies have suggested to some that future JE vaccine development must account for signi®cant antigenic variation among JE isolates. JE genome sequence data have been used to de®ne more than one ``genotype'' of JEV [7]. In addition, N assay results demonstrate that not all strains of JE are neutralized by all JE-speci®c polyclonal antisera. Speci®cally, this was the result when antisera to the Nakayama vaccine strain, which has been con®ned to laboratory propagation for more than 50 years, were used in vitro to neutralize JE isolates currently circulating in SE Asia [8]. However, in the absence of conclusive evidence to the contrary, JE vaccine development could proceed on the assumption that there is one serotype of JEV. Evidence to favor this belief is as follows: (i) secondary JE CNS infections have not been documented; JEimmune individuals appear to be protected against all circulating JE viruses. (ii) As previously mentioned, the results of the ecacy study for the Biken inactivated vaccine showed that the Nakayama-strain-only inactivated vaccine was fully as protective as the bivalent (Nakayama+Beijing) vaccine, for a population exposed to a Beijing-like virus [1]. (iii) The study cited above [8], conducted in 1991, showed that only 37% of sera from Taiwanese adults who had been immunized as children using the Nakayama-strain inactivated vaccine could neutralize the prevalent strain of JE (strain JE5) in vitro. However, other data presented in the same publication show that the Nakayamastrain vaccine dramatically reduced the incidence of disease during the prior 25-year period of its use in a childhood immunization program in Taiwan. During this time period, the circulating strains of JEV were much more closely related antigenically to strain JE5 than to the Nakayama strain, suggesting that the in vitro observation was not clinically relevant. (iv) Prospective studies of N antibody responses in recently immunized children show that the Nakayama-strain inactivated vaccine does initially induce N activity against currently circulating JEV strains [9]. It is likely that these N antibodies wane prior to Nakayamastrain-speci®c N antibodies. The observed longterm protective eect of the inactivated vaccine may therefore be related to persistence of memory T cells.
29
In addition to the epidemiologic evidence for one serotype of JEV, a recent supercomputer-based survey of 92 available JEV genome sequences revealed that the degree of divergence among the nucleotide and predicted amino-acid sequences was signi®cantly smaller than that which was observed when genome sequences of other closely related viruses (i.e. the four serotypes of dengue virus and the three serotypes of poliovirus), which exhibit real serotypic dierences, were compared. The study concluded that the computer analysis did not support the existence of distinct serotypes of JEV (Tsarev, personal communication). The available data are consistent with the hypothesis that a lifelong JE-speci®c immune response, possibly based on memory T cell activity as well as neutralizing antibodies, is stimulated by either natural JE infection or immunization with inactivated JE vaccine. This response appears to protect against CNS disease due to all strains of JEV that may arise in nature. Despite the conclusion that concerns regarding serologic variations among JEV isolates are theoretical and speculative, it would seem prudent to base future vaccine development on recently isolated wild strains, if possible, rather than on laboratory strains like Nakayama. 3.2. The N assay as a surrogate for ecacy of JE vaccines Because other ¯aviviruses, most notably the dengue viruses, co-exist with JEV in endemic areas, ¯avivirus cross-reactive serologic assays, such as conventional ELISA, HI assays, and CF assays would not be useful to monitor a vaccine response in at-risk population groups. Moreover, since ¯avivirus cross-reactive antibodies do not aord cross-species protection, the induction of an immune response measured in any of these assays could not be taken per se as evidence for protection. It is possible, however, that modi®ed forms of ELISA could be used to measure protective antibodies. For example, an IgM-capture ELISA performed on paired CSF and serum samples has been used to document acute JEV infection in ill subjects [10]. It was shown to be highly sensitive and speci®c for diagnosis. Additional data would be required to support the IgM-capture ELISA as a surrogate for ecacy, and the validation of the test would have to depend on serum responses only, for a large scale vaccine trial. Speci®city of the response for JEV would have to be proven. Alternatively, protective IgG-class antibodies could be detected by another type of modi®ed ELISA, using JE-speci®c neutralizing monoclonal antibodies [11] to compete for virus-binding sites with vaccinee's sera. Such an assay could in principle be developed as a surrogate for JE vaccine ecacy. Although these variants of ELISA have potential usefulness, novel approaches are probably not required
30
L. Marko / Vaccine 18 (2000) 26±32
in order to establish a surrogate for ecacy of JE vaccines, since there is already a large body of evidence to support use of the conventional N assay as a surrogate for protection for JE vaccines. For example: (i) there is much historical data, from Japanese studies in the 1930s and from observations made by Albert Sabin and his colleagues in the 1940s and later [12], showing that passive transfer of JE antibodies to laboratory workers who had been accidentally exposed to JEV infection was protective, as long as measurable titers of N antibody resulted. (ii) During the previously cited ecacy trial for the Biken inactivated JE vaccine [1], a cohort of 01500 subjects, randomly selected from the 1- to 14-year-old age groups, were monitored for the appearance of JE-speci®c N antibodies. The vaccine eect was seen in the much earlier age-related detection of N antibodies in the vaccinees vs those who received placebo and correlated with the age-speci®c reduction in disease incidence. (iii) Animal studies of JE infection and immunity dating at least from the early 1970s have consistently been able to demonstrate a correlation between N antibodies and protection against challenge. (Some of these studies will be speci®cally cited in the following section on animal models for JE.) JEV neutralizing antibody activity in sera of naturally exposed individuals can be detected for longer times after exposure than HI or CF antibody; in one study, N antibody was still present in sera ®ve years after a group of American citizens were removed from an endemic area. Another study showed that N antibodies induced by the Nakayama-strain inactivated vaccine were still detectable after three years in sera from children living in an endemic area. The longevity of the N antibody response would additionally facilitate monitoring of a vaccine eect, assuming antibodies induced by a novel vaccine are as long-lived. If sponsors elect to establish the N assay as a basis for an immunologic surrogate for protection aorded by novel JE vaccines, a standard method for performing the test must be chosen, validated, and used consistently thereafter. Historically, many very dierent assays have been used to measure the ``neutralization'' of JEV by antibodies. These include the direct injection of virus/antibody mixtures into the brain (``mouse i.c.'') or into the peritoneal cavity of mice (``mouse i.p.'') and tissue-culture-based assays. The tests that use mice probably require 100% neutralization of the injected virus dose, in order to protect a susceptible mouse strain from lethal outcome. Results could vary in relation to the mouse strain chosen and the age of mice at the time of the test. The tissue culture assays may vary in relation to the identity of the cell substrate chosen, the condition of the monolayer on the day of the agar overlay, and the various conditions chosen for pre-incubation of virus with antibody, etc.
Finally, the tissue culture assays can also vary in relation to the chosen endpoint, e.g. 50 vs 70 vs 90% plaque reduction. In the modern era, it seems prudent to choose a tissue-culture-based assay. A protocol with a proven ``track record'', such as the one employed to measure vaccine-induced neutralizing antibodies in the Thai ecacy study [1] is advisable. 3.3. Animal models for JEV infection A wide variety of animals have been used to study JEV infection in the laboratory setting, including infant and weanling mice, hamsters, rabbits, guinea pigs, and monkeys. Here we will limit our discussion to mice and monkeys. JEV is lethal to susceptible infant mice by all routes. As mice age, their susceptibility to JEV infection is increasingly speci®c, and weanling mice are susceptible to encephalitis by either the peritoneal or the intracerebral route of inoculation. The outcome of infection in mice will be related to both the strain of mice and the strain and passage history of the JEV preparation that is used. Monkeys generally die of meningo-encephalitis after intracerebral inoculation of virus but tend to develop only an asymptomatic viremia after peripheral infection. Pathogenicity in monkeys is also JE strain-dependent. Because of the observation that virulence of JE viruses in mice and monkeys is JE strain-dependent, both species have been used in the laboratory setting as models for mechanisms of resistance to JEV infection, and data from such studies support a correlation between neutralizing antibody titers and protection. The following are a few examples of such studies: Hammon and Sather [12] passively immunized infant mice with graded doses of hyper-immune mouse serum. They showed that the decay of virus-neutralizing activity over time in sera from these mice, as measured by the ``log neutralization index'' (LNI), was linear. The initial serum LNI was directly proportional to the amount of passively transferred hyper-immune serum. Having performed this titration, they then determined the 50%-lethal dose (LD50) of virus in passively-immunized mice vs controls that had received saline instead of hyper-immune serum. They de®ned the ``log protection index'' (LPI) as the dierence between virus LD50s in the groups of passively-immunized mice vs the LD50 in controls. Mice that had a predicted serum LNI of 2.32 (using the mouse i.c. assay) had an LPI in this study of 6.93, meaning they were about 10-million-fold more resistant to lethal challenge than controls. Mice with a predicted serum LNI of 1.68 (``barely detectable'' in their estimation) had an LPI of 2.18. Mice with a serum LNI of 00.5 were still about 10-fold more resistant to lethal outcome than controls. In a related set of experiments, Lubiniecki et al.
L. Marko / Vaccine 18 (2000) 26±32
[13] studied immunity to JEV in infant ospring of hyper-immunized mothers. They demonstrated the linear decay over time of serum LNI in infant mice with maternally-acquired JEV antibodies; between the ages of 2 and 9 weeks, the serum LNI fell from about 4.4 to about 1.5. Next, they challenged groups of these mice and control infant mice at intervals as they aged and demonstrated that survival time after challenge decreased as serum LNI activity passively decayed. Finally, they passively immunized another group of infant mice with sera from the hyper-immune mothers of the original group of infant mice and repeated the observation that protection from JEV decayed in a ®rst-order linear fashion with serum LNI. A plot of LPI vs LNI for all the immunized groups of mice showed that the data fell on the same straight line (regardless of the modality for passive immunization) with a coecient of correlation between the two parameters of 0.95 ( p < <0.01). More recently, Oya reported on a similar set of experiments (cited in [14]). Two-week-old mice were injected with serial dilutions of Nakayama-strain hyper-immune serum (seven dilutions ranging from 1:1 to 1:1000). Serum titers of N antibody were determined one day later, and mice were challenged with JEV (JaTH 160 strain) on the same day. As in the previously cited studies, the LPI for passively-immunized groups of mice declined linearly in relation to the N antibody titer prior to challenge. In this study, N antibody was undetectable in groups of mice that received r1:10 dilution of hyper-immune serum (using the mouse i.c. assay), yet protection (LPI >2 vs controls) was observed for all mice, even those that received the 1:1000 dilution of hyper-immune serum. In the past year, a model for JE disease in rhesus macaques has been further developed by investigators at WRAIR [15,16]. Animals were infected intranasally with graded doses of JEV. Infection was con®rmed by detection of JEV-speci®c IgM in CSF. Histopathology performed on CNS tissues 11±14 days after infection con®rmed that early pathologic changes were similar to those observed in humans with JEV-associated CNS disease. This was the ®rst such demonstration in an animal model system. A 50%-infectious-dose (ED50) of challenge virus was determined. Subsequently, four monkeys were immunized with the Biken inactivated JE vaccine (three doses, given at 0, 7, and 28 days), and the immunized monkeys plus four JEV-naive control monkeys were given 50 ED50 of challenge virus (equal to one ED90). All four non-immune monkeys developed JEV CNS disease, whereas only one of four vaccine-immunized monkeys became ill. Macaques from this colony and the titered challenge virus are
31
being made available to quali®ed investigators interested in JE vaccine development.
4. Summary A conventional DBPC ecacy study to establish ecacy of a new JE vaccine is probably not practical. However, it may be possible to establish the ecacy of such a product by using a surrogate. The elements for establishing a surrogate for ecacy of a new JE vaccine probably should include: (i) selection of a serologic response likely to correlate with ecacy (N antibody levels for JE vaccines). (ii) The establishment of one or more animal model systems in which to verify the protective eect of the vaccine. (iii) The demonstration that protection correlates with a particular titer of N antibody elicited by the candidate vaccine. This principle could be established both by direct immunization of animals and by passive immunization of animals using human sera from vaccine-immunized subjects. (iv) Principles elucidated in the laboratory studies could be supported by a ``head-to-head'' immunogenicity study, comparing the novel vaccine to the Biken inactivated JE vaccine, which is licensed in the US. Such a study could be conducted in a seronegative population which is not at-risk for naturally-acquired JE infection. If properly designed, safety data could be obtained in the same study.
References [1] Hoke CH, et al. Protection against Japanese encephalitis by inactivated vaccine. N Eng J Med 1988;319:608±14. [2] Szmuness W, et al. A controlled clinical trial of the ecacy of hepatitis B vaccine Heptavax B: ®nal report. Hepatology 1981;1:377±85. [3] Stevens CE, et al. Yeast recombinant hepatitis B vaccine in perinatal hepatitis B virus transmission: a preliminary report. J Infect 1986;13(Suppl. A). [4] Purcell RH, et al. Inactivated hepatitis A vaccine: active and passive immunoprophylaxis in chimpanzees. Vaccines 1992;10(Suppl 1):148±51. [5] Stapleton JT, et al. Neutralization escape mutants de®ne a dominant immunogenic neutralization site on hepatitis A virus. J Virol 1987;61:491±8. [6] Calisher CH, et al. Antigenic relationships between ¯aviviruses as determined by cross-neutralization tests with polyclonal antisera. J Gen Virol 1989;70:37±43. [7] Tsuchie H, et al. Genotypes of Japanese encephalitis virus isolated in three states in Malaysia. Am J Trop Med Hyg 1997;56:153±8. [8] Shyu WR-H, Wang Y-C, Chin C, Chen W-J. Assessment of neutralizing antibodies elicited by a vaccine (Nakayama) strain of Japanese encephalitis virus in Taiwan. Epidemiol Infect 1997;119:79±83. [9] Juang RF, et al. Neutralizing antibody response to Japanese encephalitis vaccine in children. Biken J 1983;26:25±34. [10] Solomon T, et al. Rapid diagnosis of Japanese encephalitis by
32
L. Marko / Vaccine 18 (2000) 26±32
using an immunoglobulin M dot enzyme immunoassay. J Clin Microbiol 1998;36:2030±4. [11] Kimura-Kuroda J, Yasui K. Topographical analysis of antigenic determinants on the envelope glycoprotein V3 (E) of Japanese encephalitis virus using monoclonal antibodies. J Virol 1983;45:124±32. [12] Hammon WM, Sather GE. Passive immunity for arbovirus infection I. Arti®cially induced prophylaxis in man and mouse for Japanese (B) encephalitis. Am J Trop Med Hyg 1973;22:524±33. [13] Lubniecki AS, Cypress RH, Hammon WM. Passive immunity for arbovirus infection II. Quantitative aspects of naturally and
arti®cially acquired protection in mice for Japanese (B) encephalitis. Am J Trop Med Hyg 1973;22:535±42. [14] Oya A. Japanese encephalitis vaccine. Acta Pediatrica Jpn 1988;30:175±84. [15] Raengsakulrach B, et al. An intranasal challenge model for testing Japanese encephalitis vaccines in rhesus monkeys. Am J Trop Med Hyg 1999;60:329±37. [16] Khin SAM, et al. Production of lethal infection that resembles fatal human disease by intranasal inoculation of macaques with Japanese encephalitis virus. Am J Trop Med Hyg 1999;60:338± 42.
www.elsevier.com/locate/vaccine
Discussion
Points to consider in the development of a surrogate for ecacy of novel Japanese encephalitis virus vaccines Lewis Marko 1 Laboratory of Vector-borne Virus Diseases, Center for Biologics Evaluation and Research, Food and Drug Administration, Rockville, MD20852-1448, USA
Abstract Although an eective killed virus vaccine to prevent illness due to Japanese encephalitis virus (JEV) infection exists, many authorities recognize that a safe, eective live JEV vaccine is desirable in order to reduce the cost and the number of doses of vaccine required per immunization. A large-scale clinical ecacy trail for such a vaccine would be both unethical and impractical. Therefore, a surrogate for the ecacy of JE vaccines should be established. Detection of virus-neutralizing antibodies in sera of vaccinees could constitute such a surrogate for eacy. Field studies of vaccinees in endemic areas and studies done in mice already exist to support this concept. Also, titers of virus-neutralizing antibodies are already accepted as a surrogate for the ecacy of yellow fever virus vaccines and for the ecacy of other viral vaccines as well. In developing a correlation between N antibody titers and protection from JEV infection, standard procedures must be validated and adopted for both measuring N antibodies and for testing in animals. A novel live virus vaccine could be tested in the mouse and/or the monkey model of JEV infection to establish a correlation between virus-neutralizing antibodies elicited by the vaccines and protection from encephalitis. In addition, sera of subjects receiving the novel live JEV vaccine in early clinical trials could be passively transferred to mice or monkeys in order to establish the protective immunogenicity of the vaccine in humans. A monkey model for JEV infection was recently established by scientists at WRAIR in the US. From this group, pools of JEV of known infectivity for Rhesus macaques may be obtained for testing of immunity elicited by live JE vaccine virus. Published by Elsevier Science Ltd.
1. Introduction The ecacy of a vaccine is de®ned as the relative ability of the vaccine to induce a protective immune response. Ecacy is classically measured by the results of a clinical trial in which a reduction of disease incidence in an at-risk population is directly proven. Most clinical trials designed to demonstrate vaccine ecacy are doubly blinded and placebo-controlled (DBPC). This is an essential element in the plan of such a trial, if the intent is to produce direct evidence for ecacy. A DBPC study of the ecacy of formalin-inactivated 1 The opinions expressed herein are solely those of the author and do not necessarily represent the ocial policy of the US Food and Drug Administration
0264-410X/00/$ - see front matter Published by Elsevier Science Ltd. PII: S 0 2 6 4 - 4 1 0 X ( 0 0 ) 0 0 0 3 8 - 4
Japanese encephalitis virus (JEV) vaccines was conducted in Thailand, in 1984±85 [1]. About 65,000 schoolchildren, ages 1±14, were randomly divided into three groups and immunized with two doses of vaccine derived from the Nakayama strain of JEV, or vaccine derived from a mixture of Nakayama and Beijing strains of JEV, or placebo. The Beijing strain of JEV most closely resembles strains of JE that currently circulate in Thailand and the remainder of southeast Asia, while the Nakayama strain was isolated in the 1930s and has been propagated in the laboratory since then. The children were monitored for vaccine safety and illness. The diagnosis of JEV-related illness was made by the detection of JEV-speci®c IgM in either serum or cerebro-spinal ¯uid (CSF) in children with fever. During the post-immunization surveillance period, there were 51 cases of JE per 100,000 popu-
L. Marko / Vaccine 18 (2000) 26±32
lation in the placebo group vs ®ve cases of JE per 100,000 in the vaccinated group. This gave a calculated ecacy of 91% with a 70±97% con®dence interval (CI) for both vaccinated groups vs the group that received placebo. The study also demonstrated that any antigenic dierences between the Nakayama and Beijing strains of virus used to prepare the vaccines were not consequential for ecacy; children immunized with the Nakayama-strain-only vaccine were as well protected as children who received the bivalent formulation. The results of this trial were a pivotal element in the approval process for the Nakayamastrain JE vaccine that is licensed in the US. There is no absolute standard for the level of ecacy of a vaccine that is a candidate for licensure in the US. Historically, only the ecacy of poliovirus vaccines has ever been mandated by law. The Code of Federal Regulations (CFR) stipulated that such vaccines must protect 90% of vaccinees with 95% con®dence, but this section of the CFR was deleted in 1996. A variety of dierent vaccines with a wide range of demonstrated ecacy have recently been licensed in the US. The acellular pertussis vaccines, Certiva, Tripedia, and Infanrix were shown to be 71%, 80%, and 83% ecacious, respectively. A new rotavirus vaccine, Rotashield, was shown to be 90% ecacious in the prevention of severe rotavirus gastroenteritis but only 50% ecacious in the prevention of mild signs of rotavirus infection. Two inactivated hepatitis A virus (HAV) vaccines, Havrix and Vaqta, were at or near 100% ecacy in the prevention of illness. Factors that are considered in determining whether the level of ecacy of a new vaccine is acceptable for licensure include the availability of an alternative vaccine, the ecacy and other qualities of that alternative vaccine (e.g. cost, adverse event pro®le, stability), and the prevalence and severity of the disease that is to be prevented by the new vaccine. 2. Surrogates for vaccine ecacy A classical ecacy trial of a novel vaccine may be impractical in the modern era, for all or some of the following reasons: (i) vaccines to prevent many of the major infectious diseases are already available. Often the study vaccine will not be the only vaccine with the potential to prevent disease caused by the targeted pathogen. If the alternative vaccines are already licensed, it may not be ethical to perform an ecacy trial that includes a placebo control group. This consideration currently pertains to the development and testing of new JEV vaccines. (ii) Due to a pre-existing vaccine eect, the incidence of disease may be so low that a prohibitively large study would be required to adduce a statistically signi®cant result. For example,
27
the incidence of JE clinical illness in Southeast Asia is currently estimated to be 10 cases per 100,000. Given this incidence of disease, it would require 247,000 subjects per group (vaccinees vs placebo) to demonstrate 90% ecacy with a CI similar to that which was demonstrated in the Thai study (Lachenbruch, personal communication). Such a clinical trial would be prohibitively costly and dicult to perform. (iii) It may be impossible to design a valid DBPC clinical trial to establish the ecacy of a combination vaccine in a population that may have widely variant levels of pre-existing immunity to one vs another of the targeted pathogens or in which the inclusion of a placebo group is not acceptable for the same reasons applicable to the study of monovalent vaccines. In order to circumvent the problems associated with executing a DBPC clinical trial for vaccine ecacy, regulatory authorities may permit the substitution of data establishing a ``surrogate'' for ecacy. This is a laboratory test result, usually a serologic assay, that has an accepted positive correlation with the induction by a vaccine of a protective immune response in vaccinees. The induction of a certain minimum titer of antibodies in vaccinees, typically as detected by a speci®c RIA or ELISA, a virus-neutralization (N) assay, or an assay for hemagglutination-inhibiting (HI) antibodies, is often taken as a surrogate for a protective eect of a vaccine. For example, the N assay provides a surrogate for ecacy of poliovirus vaccines; a titer of >1:8 is taken as indicative of protection against poliovirus infection. ELISAs or RIAs using HAV as antigen provide surrogates for ecacy of hepatitis A and B vaccines. In contrast, no serologic assay has been shown to correlate with the ecacy of rotavirus or pertussis vaccines. Where a surrogate for ecacy has been identi®ed, it may be sucient to demonstrate that a new vaccine stimulates a protective immune response in a large and heterogeneous cohort of seronegative subjects. One major advantage of relying upon a surrogate to show protection, as opposed to a classical ecacy trial, is that a study using a surrogate may be carried out in a population that is not at risk for the targeted disease. This simpli®es study design and reduces safety concerns. If a similar licensed vaccine already exists, one form that such a study may take is that of a random, doubly-blinded ``head-to-head'' comparison between the trial vaccine and the previously licensed product. Issues of safety as well as ecacy can be addressed in such a trial. A surrogate for ecacy may be supported by careful analysis of the results of a previously conducted clinical trial designed to demonstrate ecacy directly. For example, a controlled DBPC study was conducted in 1083 hepatitis B (HB) seronegative homosexual males to show ecacy for Heptavax, the ®rst HB vaccine to
28
L. Marko / Vaccine 18 (2000) 26±32
be approved in the US [2]. About half the subjects were immunized, and all were monitored for 24 months for the occurrence of subsequent HB infection. The overall incidence of the disease was 25.6% in the placebo group and 3.2% in the vaccinated group. Further analysis of the cases among vaccinees revealed a negative correlation between antibody titer after immunization and disease occurrence. Among 21 vaccinees who had titers R2.1 RU (Ausab Ratio Units), there were seven cases, for an incidence of 33.3%. This was essentially not dierent from the incidence in the placebo group. Among nine subjects who had titers between 2.1 RU and 10 RU, there was one case, for an incidence of 11.11%. And among 421 vaccinees who had titers r10 RU, there was also only one case, for an incidence of 0.23% and nearly 100% vaccine ecacy in this group. This result established the concept that the titer of antibody (r10 RU or mIU/ml) was a correlate of protection. Heptavax was a ``®rst-generation'' vaccine; it was composed of HBsAg particles puri®ed from the sera of infected donors. Heptavax was superceded in the US by vaccines derived via recombinant DNA technology in yeast. The immunizing antigen in these second-generation HB vaccines (Recombivax and Engerix B) was shown by sponsors to be physically identical to the antigen in Heptavax by available methodologies. However, when the ®rst of these vaccines (Recombivax) was under review, a small-scale additional ecacy study (in infants borne of mothers who were chronic carriers of HB [3]) was required as a supplement to an immunogenicity study demonstrating that Recombivax induced antibody titers r10 mIU/ml in an Ausab-like assay in nearly 100% of 1500 adult subjects who received the recommended vaccine dosage. Thus, a surrogate for ecacy based on data obtained with a vaccine derived using one speci®c technology will not automatically be acceptable for a vaccine derived using an entirely dierent one, though the target disease and even the antigen content of the vaccine may be the same in both cases. After the Recombivax approval process, FDA required no formal ecacy trial in support of approval of the second yeast-derived HB vaccine, Engerix B; the immune correlate of protection was accepted. Animal studies can be used to provide support for a surrogate for ecacy of a vaccine, especially where there exists an animal model that mimics pathogenesis of the disease in humans. Evidence of this kind was available in support of the use of RIA or ELISA results as a surrogate for ecacy of hepatitis A (HA) vaccines. Chimps were used as a model for the disease in humans. It was initially shown that chimps passively immunized with pooled hyper-immune serum from HA-infected humans were protected from challenge. Next, chimps were challenged at dierent time inter-
vals after passive immunization. As antibody titers waned, these chimps remained protected as long as they had serum antibody levels that were at or above detectability using a speci®c RIA (HAVAB or modi®ed HAVAB, available in kit form from Abbott Labs) or by a speci®c ELISA, ``ELISA-Ig'' (Smithkline-Beecham Biologicals) [4]. Both are competition assays with identical cut-os, using puri®ed virions as antigen. In additional studies, it was shown that the antibodies in human sera detected in the ELISA-Ig could be completely blocked by a mixture of two monoclonal antibodies that recognized non-overlapping epitopes with virus-neutralizing activity [5]. This established a correlation between ELISA antibodies and N antibodies elicited by HA infection. The correlation was borne out by studies of sera from subjects who had naturally acquired HA infection or who had been immunized with the inactivated HA vaccine, Havrix. In summary, animal and laboratory studies established that ELISAmeasured antibodies had N activity and that vaccineinduced antibodies were qualitatively similar to those resulting from natural infection. Finally, a large-scale clinical trial to show ecacy of Havrix was conducted among more than 40,000 schoolchildren in Thailand. The results of this study were accepted as evidence for the ecacy of Havrix in support of its approval in the US. The overall outcome of the approval process for Havrix (after FDA review of the formal ecacy trial and the earlier chimp and serologic studies) supported the proposition that antibody detected in the ELISA-Ig [and more recently, the ``Enzymun'' assay (Boehringer-Mannheim]) indicates the presence of protective immunity in a vaccinee who has received an inactivated whole virus HA vaccine. This principle can now be applied in the approval process for similar HA vaccines.
3. Surrogates for ecacy of JE vaccines The past history suggests a plausible approach to establishing a surrogate for the ecacy of new JE vaccines. Elements of this process could include: (i) the selection of one or more likely immunologic surrogates for protection. (ii) The selection of an animal model for JE pathogenesis. (iii) Performing passive protection experiments in the selected animal model, when possible using human antibodies from naturally-infected subjects and from vaccinees. (iv) Correlation of the results of such studies with antibody levels as determined by the projected surrogate assay. (v) Conduct of a ``head-to-head'' immunogenicity (and safety) trial, using an existing JE vaccine, for example the USlicensed inactivated JEV vaccine (Biken/Connaught Labs), as a comparator.
L. Marko / Vaccine 18 (2000) 26±32
3.1. Evidence for one serotype of JEV Flavivirus infections elicit neutralizing and non-neutralizing antibodies that may have serotype, species, subgroup or group speci®city. Flavivirus antibodies have been commonly measured by conventional ELISAs, HI assays, assays for complement-®xing (CF) antibody, and the N assay. As normally performed, ELISA and HI and CF assays are highly cross-reactive, whereas the N assay provides the basis for speciation of ¯aviviruses [6]. Laboratory studies have suggested to some that future JE vaccine development must account for signi®cant antigenic variation among JE isolates. JE genome sequence data have been used to de®ne more than one ``genotype'' of JEV [7]. In addition, N assay results demonstrate that not all strains of JE are neutralized by all JE-speci®c polyclonal antisera. Speci®cally, this was the result when antisera to the Nakayama vaccine strain, which has been con®ned to laboratory propagation for more than 50 years, were used in vitro to neutralize JE isolates currently circulating in SE Asia [8]. However, in the absence of conclusive evidence to the contrary, JE vaccine development could proceed on the assumption that there is one serotype of JEV. Evidence to favor this belief is as follows: (i) secondary JE CNS infections have not been documented; JEimmune individuals appear to be protected against all circulating JE viruses. (ii) As previously mentioned, the results of the ecacy study for the Biken inactivated vaccine showed that the Nakayama-strain-only inactivated vaccine was fully as protective as the bivalent (Nakayama+Beijing) vaccine, for a population exposed to a Beijing-like virus [1]. (iii) The study cited above [8], conducted in 1991, showed that only 37% of sera from Taiwanese adults who had been immunized as children using the Nakayama-strain inactivated vaccine could neutralize the prevalent strain of JE (strain JE5) in vitro. However, other data presented in the same publication show that the Nakayamastrain vaccine dramatically reduced the incidence of disease during the prior 25-year period of its use in a childhood immunization program in Taiwan. During this time period, the circulating strains of JEV were much more closely related antigenically to strain JE5 than to the Nakayama strain, suggesting that the in vitro observation was not clinically relevant. (iv) Prospective studies of N antibody responses in recently immunized children show that the Nakayama-strain inactivated vaccine does initially induce N activity against currently circulating JEV strains [9]. It is likely that these N antibodies wane prior to Nakayamastrain-speci®c N antibodies. The observed longterm protective eect of the inactivated vaccine may therefore be related to persistence of memory T cells.
29
In addition to the epidemiologic evidence for one serotype of JEV, a recent supercomputer-based survey of 92 available JEV genome sequences revealed that the degree of divergence among the nucleotide and predicted amino-acid sequences was signi®cantly smaller than that which was observed when genome sequences of other closely related viruses (i.e. the four serotypes of dengue virus and the three serotypes of poliovirus), which exhibit real serotypic dierences, were compared. The study concluded that the computer analysis did not support the existence of distinct serotypes of JEV (Tsarev, personal communication). The available data are consistent with the hypothesis that a lifelong JE-speci®c immune response, possibly based on memory T cell activity as well as neutralizing antibodies, is stimulated by either natural JE infection or immunization with inactivated JE vaccine. This response appears to protect against CNS disease due to all strains of JEV that may arise in nature. Despite the conclusion that concerns regarding serologic variations among JEV isolates are theoretical and speculative, it would seem prudent to base future vaccine development on recently isolated wild strains, if possible, rather than on laboratory strains like Nakayama. 3.2. The N assay as a surrogate for ecacy of JE vaccines Because other ¯aviviruses, most notably the dengue viruses, co-exist with JEV in endemic areas, ¯avivirus cross-reactive serologic assays, such as conventional ELISA, HI assays, and CF assays would not be useful to monitor a vaccine response in at-risk population groups. Moreover, since ¯avivirus cross-reactive antibodies do not aord cross-species protection, the induction of an immune response measured in any of these assays could not be taken per se as evidence for protection. It is possible, however, that modi®ed forms of ELISA could be used to measure protective antibodies. For example, an IgM-capture ELISA performed on paired CSF and serum samples has been used to document acute JEV infection in ill subjects [10]. It was shown to be highly sensitive and speci®c for diagnosis. Additional data would be required to support the IgM-capture ELISA as a surrogate for ecacy, and the validation of the test would have to depend on serum responses only, for a large scale vaccine trial. Speci®city of the response for JEV would have to be proven. Alternatively, protective IgG-class antibodies could be detected by another type of modi®ed ELISA, using JE-speci®c neutralizing monoclonal antibodies [11] to compete for virus-binding sites with vaccinee's sera. Such an assay could in principle be developed as a surrogate for JE vaccine ecacy. Although these variants of ELISA have potential usefulness, novel approaches are probably not required
30
L. Marko / Vaccine 18 (2000) 26±32
in order to establish a surrogate for ecacy of JE vaccines, since there is already a large body of evidence to support use of the conventional N assay as a surrogate for protection for JE vaccines. For example: (i) there is much historical data, from Japanese studies in the 1930s and from observations made by Albert Sabin and his colleagues in the 1940s and later [12], showing that passive transfer of JE antibodies to laboratory workers who had been accidentally exposed to JEV infection was protective, as long as measurable titers of N antibody resulted. (ii) During the previously cited ecacy trial for the Biken inactivated JE vaccine [1], a cohort of 01500 subjects, randomly selected from the 1- to 14-year-old age groups, were monitored for the appearance of JE-speci®c N antibodies. The vaccine eect was seen in the much earlier age-related detection of N antibodies in the vaccinees vs those who received placebo and correlated with the age-speci®c reduction in disease incidence. (iii) Animal studies of JE infection and immunity dating at least from the early 1970s have consistently been able to demonstrate a correlation between N antibodies and protection against challenge. (Some of these studies will be speci®cally cited in the following section on animal models for JE.) JEV neutralizing antibody activity in sera of naturally exposed individuals can be detected for longer times after exposure than HI or CF antibody; in one study, N antibody was still present in sera ®ve years after a group of American citizens were removed from an endemic area. Another study showed that N antibodies induced by the Nakayama-strain inactivated vaccine were still detectable after three years in sera from children living in an endemic area. The longevity of the N antibody response would additionally facilitate monitoring of a vaccine eect, assuming antibodies induced by a novel vaccine are as long-lived. If sponsors elect to establish the N assay as a basis for an immunologic surrogate for protection aorded by novel JE vaccines, a standard method for performing the test must be chosen, validated, and used consistently thereafter. Historically, many very dierent assays have been used to measure the ``neutralization'' of JEV by antibodies. These include the direct injection of virus/antibody mixtures into the brain (``mouse i.c.'') or into the peritoneal cavity of mice (``mouse i.p.'') and tissue-culture-based assays. The tests that use mice probably require 100% neutralization of the injected virus dose, in order to protect a susceptible mouse strain from lethal outcome. Results could vary in relation to the mouse strain chosen and the age of mice at the time of the test. The tissue culture assays may vary in relation to the identity of the cell substrate chosen, the condition of the monolayer on the day of the agar overlay, and the various conditions chosen for pre-incubation of virus with antibody, etc.
Finally, the tissue culture assays can also vary in relation to the chosen endpoint, e.g. 50 vs 70 vs 90% plaque reduction. In the modern era, it seems prudent to choose a tissue-culture-based assay. A protocol with a proven ``track record'', such as the one employed to measure vaccine-induced neutralizing antibodies in the Thai ecacy study [1] is advisable. 3.3. Animal models for JEV infection A wide variety of animals have been used to study JEV infection in the laboratory setting, including infant and weanling mice, hamsters, rabbits, guinea pigs, and monkeys. Here we will limit our discussion to mice and monkeys. JEV is lethal to susceptible infant mice by all routes. As mice age, their susceptibility to JEV infection is increasingly speci®c, and weanling mice are susceptible to encephalitis by either the peritoneal or the intracerebral route of inoculation. The outcome of infection in mice will be related to both the strain of mice and the strain and passage history of the JEV preparation that is used. Monkeys generally die of meningo-encephalitis after intracerebral inoculation of virus but tend to develop only an asymptomatic viremia after peripheral infection. Pathogenicity in monkeys is also JE strain-dependent. Because of the observation that virulence of JE viruses in mice and monkeys is JE strain-dependent, both species have been used in the laboratory setting as models for mechanisms of resistance to JEV infection, and data from such studies support a correlation between neutralizing antibody titers and protection. The following are a few examples of such studies: Hammon and Sather [12] passively immunized infant mice with graded doses of hyper-immune mouse serum. They showed that the decay of virus-neutralizing activity over time in sera from these mice, as measured by the ``log neutralization index'' (LNI), was linear. The initial serum LNI was directly proportional to the amount of passively transferred hyper-immune serum. Having performed this titration, they then determined the 50%-lethal dose (LD50) of virus in passively-immunized mice vs controls that had received saline instead of hyper-immune serum. They de®ned the ``log protection index'' (LPI) as the dierence between virus LD50s in the groups of passively-immunized mice vs the LD50 in controls. Mice that had a predicted serum LNI of 2.32 (using the mouse i.c. assay) had an LPI in this study of 6.93, meaning they were about 10-million-fold more resistant to lethal challenge than controls. Mice with a predicted serum LNI of 1.68 (``barely detectable'' in their estimation) had an LPI of 2.18. Mice with a serum LNI of 00.5 were still about 10-fold more resistant to lethal outcome than controls. In a related set of experiments, Lubiniecki et al.
L. Marko / Vaccine 18 (2000) 26±32
[13] studied immunity to JEV in infant ospring of hyper-immunized mothers. They demonstrated the linear decay over time of serum LNI in infant mice with maternally-acquired JEV antibodies; between the ages of 2 and 9 weeks, the serum LNI fell from about 4.4 to about 1.5. Next, they challenged groups of these mice and control infant mice at intervals as they aged and demonstrated that survival time after challenge decreased as serum LNI activity passively decayed. Finally, they passively immunized another group of infant mice with sera from the hyper-immune mothers of the original group of infant mice and repeated the observation that protection from JEV decayed in a ®rst-order linear fashion with serum LNI. A plot of LPI vs LNI for all the immunized groups of mice showed that the data fell on the same straight line (regardless of the modality for passive immunization) with a coecient of correlation between the two parameters of 0.95 ( p < <0.01). More recently, Oya reported on a similar set of experiments (cited in [14]). Two-week-old mice were injected with serial dilutions of Nakayama-strain hyper-immune serum (seven dilutions ranging from 1:1 to 1:1000). Serum titers of N antibody were determined one day later, and mice were challenged with JEV (JaTH 160 strain) on the same day. As in the previously cited studies, the LPI for passively-immunized groups of mice declined linearly in relation to the N antibody titer prior to challenge. In this study, N antibody was undetectable in groups of mice that received r1:10 dilution of hyper-immune serum (using the mouse i.c. assay), yet protection (LPI >2 vs controls) was observed for all mice, even those that received the 1:1000 dilution of hyper-immune serum. In the past year, a model for JE disease in rhesus macaques has been further developed by investigators at WRAIR [15,16]. Animals were infected intranasally with graded doses of JEV. Infection was con®rmed by detection of JEV-speci®c IgM in CSF. Histopathology performed on CNS tissues 11±14 days after infection con®rmed that early pathologic changes were similar to those observed in humans with JEV-associated CNS disease. This was the ®rst such demonstration in an animal model system. A 50%-infectious-dose (ED50) of challenge virus was determined. Subsequently, four monkeys were immunized with the Biken inactivated JE vaccine (three doses, given at 0, 7, and 28 days), and the immunized monkeys plus four JEV-naive control monkeys were given 50 ED50 of challenge virus (equal to one ED90). All four non-immune monkeys developed JEV CNS disease, whereas only one of four vaccine-immunized monkeys became ill. Macaques from this colony and the titered challenge virus are
31
being made available to quali®ed investigators interested in JE vaccine development.
4. Summary A conventional DBPC ecacy study to establish ecacy of a new JE vaccine is probably not practical. However, it may be possible to establish the ecacy of such a product by using a surrogate. The elements for establishing a surrogate for ecacy of a new JE vaccine probably should include: (i) selection of a serologic response likely to correlate with ecacy (N antibody levels for JE vaccines). (ii) The establishment of one or more animal model systems in which to verify the protective eect of the vaccine. (iii) The demonstration that protection correlates with a particular titer of N antibody elicited by the candidate vaccine. This principle could be established both by direct immunization of animals and by passive immunization of animals using human sera from vaccine-immunized subjects. (iv) Principles elucidated in the laboratory studies could be supported by a ``head-to-head'' immunogenicity study, comparing the novel vaccine to the Biken inactivated JE vaccine, which is licensed in the US. Such a study could be conducted in a seronegative population which is not at-risk for naturally-acquired JE infection. If properly designed, safety data could be obtained in the same study.
References [1] Hoke CH, et al. Protection against Japanese encephalitis by inactivated vaccine. N Eng J Med 1988;319:608±14. [2] Szmuness W, et al. A controlled clinical trial of the ecacy of hepatitis B vaccine Heptavax B: ®nal report. Hepatology 1981;1:377±85. [3] Stevens CE, et al. Yeast recombinant hepatitis B vaccine in perinatal hepatitis B virus transmission: a preliminary report. J Infect 1986;13(Suppl. A). [4] Purcell RH, et al. Inactivated hepatitis A vaccine: active and passive immunoprophylaxis in chimpanzees. Vaccines 1992;10(Suppl 1):148±51. [5] Stapleton JT, et al. Neutralization escape mutants de®ne a dominant immunogenic neutralization site on hepatitis A virus. J Virol 1987;61:491±8. [6] Calisher CH, et al. Antigenic relationships between ¯aviviruses as determined by cross-neutralization tests with polyclonal antisera. J Gen Virol 1989;70:37±43. [7] Tsuchie H, et al. Genotypes of Japanese encephalitis virus isolated in three states in Malaysia. Am J Trop Med Hyg 1997;56:153±8. [8] Shyu WR-H, Wang Y-C, Chin C, Chen W-J. Assessment of neutralizing antibodies elicited by a vaccine (Nakayama) strain of Japanese encephalitis virus in Taiwan. Epidemiol Infect 1997;119:79±83. [9] Juang RF, et al. Neutralizing antibody response to Japanese encephalitis vaccine in children. Biken J 1983;26:25±34. [10] Solomon T, et al. Rapid diagnosis of Japanese encephalitis by
32
L. Marko / Vaccine 18 (2000) 26±32
using an immunoglobulin M dot enzyme immunoassay. J Clin Microbiol 1998;36:2030±4. [11] Kimura-Kuroda J, Yasui K. Topographical analysis of antigenic determinants on the envelope glycoprotein V3 (E) of Japanese encephalitis virus using monoclonal antibodies. J Virol 1983;45:124±32. [12] Hammon WM, Sather GE. Passive immunity for arbovirus infection I. Arti®cially induced prophylaxis in man and mouse for Japanese (B) encephalitis. Am J Trop Med Hyg 1973;22:524±33. [13] Lubniecki AS, Cypress RH, Hammon WM. Passive immunity for arbovirus infection II. Quantitative aspects of naturally and
arti®cially acquired protection in mice for Japanese (B) encephalitis. Am J Trop Med Hyg 1973;22:535±42. [14] Oya A. Japanese encephalitis vaccine. Acta Pediatrica Jpn 1988;30:175±84. [15] Raengsakulrach B, et al. An intranasal challenge model for testing Japanese encephalitis vaccines in rhesus monkeys. Am J Trop Med Hyg 1999;60:329±37. [16] Khin SAM, et al. Production of lethal infection that resembles fatal human disease by intranasal inoculation of macaques with Japanese encephalitis virus. Am J Trop Med Hyg 1999;60:338± 42.