INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Neisseria meningitidis is an important cause of morbidity and mortality and is a leading cause of bacterial meningitis and septicemia in children and young adults. Since 1997, meningococcal disease caused by meningococcal serogroup B (MenB) has been considered endemic in Brazil, although small outbreaks caused by MenC have been reported in some states.

A large epidemic of meningococcal disease due to MenB has been occurring in greater São Paulo (GSP), São Paulo State, since 1988 (27). In 1990, the incidence of meningococcal disease reached 6.2 per 100,000 inhabitants, and 50% of the isolated strains were serogroup B, serotype 4,7, serosubtype P1.19,15 (B:4,7:P1.19,15). In 1989 and 1990, two doses of the Cuban-produced outer membrane vesicle (OMV)-based vaccine (B:4,7:P1.19,15) were given to approximately 2.4 million Brazilian children from 3 months to 6 years of age (92% of the children in the target age range). Although most of the isolates in GSP matched the phenotype of the vaccine-type strain, and despite a high level of population coverage, the administration of that vaccine had little or no measurable effect on this epidemic (7). The incidence of serogroup B meningococcal disease in 1991, a year after the vaccination, was reduced only by 3% (7). In 1996, the incidence of meningococcal disease reached 7.8 per 100,000 inhabitants per year, the highest since the 1970s, when large epidemics were caused by serogroup A or C (20, 22). Sixty-one percent of the serogrouped strains in 1996 were MenB. Since then, the actual trend in the rate of meningococcal disease in GSP has been declining, reaching 5.5 per 100,000 inhabitants per year in 1999 with the same proportion of cases due to MenB as in 1996.

Vaccines based on the capsular polysaccharide (CPS) of serogroups A, C, W135, and Y have been developed and have been shown to control outbreaks or epidemics of meningococcal disease (11, 12, 23). However, serogroup B CPS is poorly immunogenic and induces only a transient antibody response of a predominantly immunoglobulin M isotype (3, 14, 42). Other than capsular antigens, outer membrane proteins (OMPs), primarily class 1 to 5, have been studied as potential vaccine candidates (32). PorA protein (class 1 protein, the serosubtype antigen) and the mutually exclusive PorB protein (class 2 and 3 proteins, the serotype antigen) have been most extensively studied. Based on the immunologic hypervariability of these antigens, meningococcal strains are subdivided into a number of serotypes and serosubtypes (1, 2, 10).

A two-dimensional structural model containing eight exposed surface loops (loops I to VIII) has been predicted for meningococcal PorA protein (35). Most variability between PorA proteins resides in the variable regions (VRs), VR1 and VR2, which correspond to loops I and IV, respectively (15, 17, 28, 35). A three-dimensional structural model for meningococcal PorA has recently been predicted that provides information on the spatial relationships between VRs in the PorA trimer (8). Serosubtype-defining monoclonal antibodies (MAbs) react with peptide epitopes located in those loops; therefore, serosubtyping has been used to identify those PorA VR epitopes.

Immunization of volunteers with experimental vaccines based on OMVs demonstrated that bactericidal polyclonal antibodies were mainly directed against the PorA protein, and the presence of these antibodies has been generally accepted as a marker for protection against infection by strains with the same serosubtype(s) as those of the vaccine strain (4, 13, 21, 25, 26, 38, 39). Over the past decade, single- and multivalent vaccines with different compositions of PorA epitopes have been developed based on the most prevalent serosubtypes in distinct geographic areas and used in clinical trials (5, 6, 21, 24, 31, 36, 38). Because the serosubtype prevalence of MenB may change over time, continuous monitoring of the distribution of the circulating serosubtypes in a particular region is necessary to adapt the vaccine composition to accommodate the changes in prevalence of dominant serosubtypes.

We previously demonstrated that the current panel of serosubtype-defining MAbs underestimates PorA VR diversity by at least 50% (28). Therefore, the serosubtyping system is not comprehensive, and many isolates remain non-serosubtypeable (NSST) or partially serosubtypeable (PSST), which hampers the use of this method as a predictor of serosubtypes for inclusion in any new MenB vaccine (33). The application of direct porA sequence determination permits characterization of meningococcal PorA VRs by predicting amino acid sequence (PorA VR typing) as an alternative to serosubtyping (9, 28).

In this study, we determined the prevalence of serosubtypes of MenB isolated in several Brazilian states during 1997 and 1998. This information may be used to define the serosubtype antigen composition for the Brazilian MenB vaccine under development or any other OMV-based vaccine to be used in Brazil. We also investigated the extent of PorA VR sequence variation among the most prevalent serosubtypes; this information may be used to select the vaccine strains with identical PorA VR sequences, as found in the most prevalent serosubtypes. In this study, we address the implications of those results for designing an OMV-based serosubtype vaccine for prevention of serogroup B meningococcal disease in Brazil.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Meningococcal disease case definition. Cases of meningococcal disease are voluntarily reported, and the following criteria serve as the basis for case definition: (i) N. meningitidis isolated from cerebrospinal fluid (CFS) or blood, (ii) meningococcal antigens demonstrated in CSF or serum by counterimmunoelectrophoresis or latex agglutination, (iii) gram-negative diplococci identified by Gram stain of CFS or serum, (iv) abnormal CSF cytopathologic findings in a patient who has acute hemorrhagic skin lesions, or (v) symptoms of bacterial meningitis and hemorrhagic skin lesions with either normal CSF or no CSF data available.

Epidemiologic data. The epidemiologic data on meningococcal disease in Brazil were collected and analyzed by the Brazilian Ministry of Health-National Foundation of Health, National Center for Epidemiology, Respiratory Diseases Surveillance Department, Brasilia, and the Center of Epidemiological Surveillance Alexandre Vranjaque, São Paulo. The population data were obtained from the Brazilian Institute for Geography and Statistics (IBGE; http://www.ibge.gov.br/) and reflect the estimates for 1996.

N. meningitidis strains. Meningococcal strains recovered from blood or CSF of patients with meningococcal disease in several Brazilian states are forwarded to IAL, the National Reference Center for N. meningitidis in São Paulo, for identification and molecular characterization. In this study, we analyzed 1,297 MenB strains from the IAL collection, isolated in 1997 and 1998, which represent all available MenB strains isolated during that period (Table 2). The states from which these isolates were recovered are adjacent to each other from the south to the north region of the east coast of Brazil, and together they comprised 73% of the entire Brazilian population in 1996. 

Serotyping. All 1,297 MenB strains were serosubtyped by dot blotting of whole-cell suspensions as previously described (33, 40). A fully serosubtyped (FSST) strain has both VR epitopes characterized by MAbs, while a PSST strain has only one VR (VR1 or -2) characterized by MAb.

PorA VR typing. Based on the serosubtyping results from the 1,297 MenB strains, two subsets totaling 164 strains were selected to be further analyzed by PorA VR typing. The general approach used in selection of strains was to type all PSST and NSST strains and at least 50% of statistically selected FSST strains. Consequently, the first subset represented 124 strains isolated in GSP (referred to as GSP strains), which includes all PSST and NSST strains and 57% of the FSST strains (Table 2). The second subset contained 40 MenB serosubtype P1.7,16 strains (referred to as southern strains) isolated in the three southern states of Paraná (15 strains), Santa Catarina (13 strains), and Rio Grande do Sul (12 strains), where P1.7,16 was identified as the most prevalent serosubtype. PCR primers P14 (28) and P22 (15) were used to amplify the full-length porA gene from all isolates as previously described (28). For PorA typing, only two regions of the porA gene, which include VR1 and VR2, were sequenced (29). In this study, we have used a new PorA VR typing system nomenclature for PorA VR amino acid classification and designation of the PorA family variants as recently described (29). For full-length sequencing of the porA gene, 14 primers were used as previously described (28). The sequences of the porA genes obtained during the study have been submitted to the GenBank database.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Serosubtyping of strains. (i) Serosubtyping of isolates from 10 Brazilian states. A total of 1,297 MenB strains from 10 Brazilian states were serosubtyped by dot blotting of whole-cell suspensions: 1,116 (86%) were FSST, 140 (11%) were PSST, and 41 (3%) were NSST. P1.19,15 was the most prevalent serosubtype in nine states and the Federal District, with prevalence varying in a range from 38% in Santa Catarina to 87% in the Federal District, but only 14% in Rio Grande do Sul. There was a correlation between serosubtype P1.7,16 and a specific geographic area, because 74% (40 of 54) of all P1.7,16 strains were collected in the contiguous southern states of Paraná (28%), Santa Catarina (24%), and Rio Grande do Sul (22%) (Table 2). Strains of serosubtype P1.7,16 were absent or rarely seen in most other states. The distribution of the most prevalent serosubtypes in Brazil is presented in Table 2. The serosubtyping results for 199 MenB strains isolated in GSP during the same period were similar to those obtained for the entire collection of strains: 175 (88%) were FSST, 19 (10%) were PSST, and 5 (2%) were NSST (Table 3). Serosubtype P1.19,15 was also the most prevalent in GSP and was responsible for 76% of all serogroup B cases in this period. The distribution of the most prevalent serosubtypes in all geographic areas studied is presented in Table 2.

(ii) Serotyping of vaccine study strains. The eight B:4:P1.15 vaccine study strains all had the identical serosubtypeP1.19,15.

PorA VR typing of strains from GSP. One hundred twenty-four (62%) of the 199 MenB strains isolated in GSP were PorA VR typed: all 19 PSST, all 5 NSST, and 100 (57%) of the 175 FSST strains (77 of the 152 P1.19,15 strains) and all strains with other epitope combinations (12 P1.7,1; 6 P1.22-1,14; 3 P1.7,16; 1 P1.5,10; and 1 P1.7,15).

PorA typing of strains from the southern states. The 40 strains representing all isolated MenB P1.7,16 in the southern states of Paraná, Santa Catarina, and Rio Grande do Sul, where they account for 74% of all identified serosubtypes, were PorA VR typed. No sequence variation was detected; all 40 strains had identical PorA type P1.7,16.

PorA typing of vaccine study strains. Finally, no sequence variation was detected in the eight B:4:P1.15 strains that had been used in a previous study of specificity of bactericidal antibody response in Brazilian children; all had identical PorA type P1.19,15.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The high prevalence of only three serosubtypes of MenB in 10 Brazilian states is encouraging for the use of an OMV-based vaccine (Table 2). A bivalent OMV-based vaccine prepared with serosubtypes P1.19,15 and P1.7,1 is already under phase I evaluation in Brazil. Based on our results, although this vaccine would target 77% of the meningococcal disease caused by MenB in the country, the range of strains targeted in various states would vary from 97% in Sergipe State and the Federal District to 14% in Rio Grande do Sul State. A trivalent vaccine prepared with the three most prevalent serosubtypes, P1.19,15, P1.7,1, and P1.7,16, would target 81% of the MenB strains isolated in Brazil and would especially increase the coverage in the southern states of Paraná, Santa Catarina, and Rio Grande do Sul to 73, 71, and 55%, respectively. A quite different situation is encountered in several other countries, particularly in the United States, where much greater diversity of serosubtypes has been identified (29).

Although serotyping has been used to identify prevalent antigens for MenB vaccine composition, serosubtype-defining MAbs are able to differentiate among prototypes, but not their variant sequences (28). PorA VR typing allows identification of the entire peptide region where the serosubtype epitope is located; thus, complete identification and discrimination among the variants can be obtained. This additional discrimination that PorA VR typing provides can explain the lack of MAb reactivity for some variants of PorA VR families. However, full elucidation is currently lacking regarding how much of the immunologic protection elicited after vaccination with any OMV-based MenB vaccine would be serosubtype or PorA type specific. Two studies indicate that the functional antibodies elicited after vaccination with serosubtype-based OMV vaccines exhibit low cross-reactivity with other serosubtypes not present in the vaccine preparation (21, 37). Together, these facts make PorA VR typing an attractive aid in determining meningococcal antigen vaccine composition and, subsequently, an important component in assessment of vaccine efficacy and elucidation of the mechanisms responsible for vaccine failures.

Based on the prevalence in GSP determined only by the phenotypic method of serosubtyping, a trivalent vaccine (serosubtypes P1.19,15, P1.7,1, and P1.7,16) used in this area would target approximately 84% of the meningococcal disease caused by MenB (Table 2). However, PorA VR typing of a random subset of the serosubtype P1.19,15 strains indicates that these are actually a mixture of prototype and variants of this family (Table 4). Assuming that there is no cross-protection among family members, the estimated vaccine coverage would be slightly reduced. Of the serosubtype P1.19,15 strains that were sequenced, 91% had the prototype sequence (Table 4). By extrapolating this proportion to the total number of serosubtype P1.19,15 strains isolated in the GSP area, the trivalent vaccine would target 77% of the strains. All serosubtype P1.7,1 and P1.7,16 strains have the same P1.7-1,1 variant sequence and P1.7,16 prototype sequence, respectively. If a trivalent vaccine composed of PorA types P1.19,15, P1.7-1,1, and P1.7,16 were used, there would be no additional reduction in our estimate of coverage. We did not sequence the porA gene from P1.19,15 or P1.7,1 strains from other states; however, considering the homogeneity of PorA types found in P1.19,15 and P1.7,1 strains isolated in GSP, it is likely that they would not be more diverse in those states. Further studies are needed to confirm this hypothesis.

Two doses of a vaccine derived from a Cuban epidemic strain (B:4,7:P1.19,15) of the same serotype, serosubtype, and PorA type as those of the most prevalent strain in GSP were administered to about 2.4 million children in GSP during 1989 to 1990 (7). The specificity of bactericidal antibodies in those children was analyzed by using different strains, including nine locally isolated B:4:P1.15 strains, as well as the vaccine strain CU385 and mutant strains lacking PorA, class 5 protein, or both (18, 19). Following the vaccination with the Cuban strain, only 26 and 55% of the vaccinees had at least a fourfold increase in bactericidal titers against the local B:4:P1.15 strain (N44/89) and the vaccine strain (CU385), respectively (19). The results also indicated that PorA and class 5 proteins were the main target in 75% of the individuals; bactericidal activity against a double-mutant strain lacking PorA and class 5 proteins declined significantly. The remaining 25% vaccinees had strong bactericidal activity against the double mutant strains as well as against the vaccine strain. A much higher percentage (68%) of individuals with bactericidal antibodies to a mutant strain lacking both PorA and class 5 protein was observed among the Norwegian adult vaccinees, who received three doses of the OMV-based vaccine of strain H44/76 (25). These findings suggest that bactericidal antibodies in a substantial proportion of the population may recognize a yet unidentified non-PorA OMV component or components.

Several studies have addressed the issue of bactericidal antibody levels elicited after vaccination with OMV-based MenB vaccine. Usually, postvaccination sera are tested against the homologous vaccine strain(s) and a local strain with the same serotype and serosubtype as those of the vaccine strain, and sometimes with variants obtained from the vaccine strains (5, 21, 24, 25, 31, 38). A fourfold increase in bactericidal antibodies against the homologous strain is generally accepted as a protective level against infection (4). Therefore, it is believed that bactericidal antibodies elicited by an OMV-based vaccine would kill circulating strains of the same serosubtype as the vaccine strain.

Milagres et al. (19) investigated this concept by using a set of 14 selected sera in which bactericidal activity against the vaccine strain and a local B:4:P1.15 strain (N44/89) was previously detected and characterized. However, different from other studies, the authors evaluated the presence of bactericidal activity against eight additional B:4:P1.15 strains isolated in the same area (vaccine study strains). The results showed that none of the 14 sera could kill all eight B:4:P1.15 strains; therefore, no individual could be considered completely immune to infection by all strains of this phenotype. Several hypotheses could explain these differences in bactericidal sensitivity, such as lower levels of OMP expression, differences in the amount of CPS, differences in class 5 protein expression, and even the presence of different lipooligosaccharides. Nevertheless, these results suggest that the assumption that bactericidal antibodies elicited by an OMV-based vaccine would kill circulating strains of the same serosubtype as the vaccine strain may not always be correct and that the use of bactericidal methods to predict protection needs to be reevaluated.

More recently, studies have shown that a decrease or lack of bactericidal activity against strains of the same serosubtype may be related to variations in PorA VR sequences (16). By serotyping with a larger set of serotyping MAbs and PorA VR typing, we showed that all eight of those original strains used by Milagres et al. 1998 are B:4,7:P1.19,15 and PorA type P1.19,15, identical to the Cuban vaccine strain CU385 and strain N44/89 originally used for bactericidal assays. Therefore, serotype and serosubtype differences or PorA VR type variation cannot explain the differences in bactericidal activity found among those 14 sera and the eight different P1.19,15 MenB strains previously used for bactericidal tests.

An OMV-based vaccine is more likely to kill strains of the same serosubtype or PorA type as the vaccine strain; however, according to these data, such a vaccine can be considered neither serosubtype nor PorA type specific. PorA protein is the main protective component in an OMV-based vaccine, but the role of minor OMPs and other nonprotein OMV components present in the vaccine cannot be ignored.

Even though the strains analyzed in this study were not collected through an active laboratory-based surveillance, they indeed represent all isolates submitted through the passive laboratory surveillance and are currently the best available resource for obtaining information about the characteristics of MenB strains in Brazil. According to our results, a trivalent OMV-based vaccine prepared with PorA types P1.19,15, P1.7-1,1, and P1.7,16 may be appropriate to control meningococcal disease in most of the Brazilian states. The coverage by using this vaccine would be easy to estimate, but the efficacy of such a vaccine in that population is still difficult to predict, since several other factors that are not well understood affect the level of protection.