Genomic Surveillance of 4CMenB Vaccine Antigenic Variants among Disease-Causing Neisseria meningitidis Isolates, United Kingdom, 2010–2016

In September 2015, 4CMenB meningococcal vaccine was introduced into the United Kingdom infant immunization program without phase 3 trial information. Understanding the effect of this program requires enhanced surveillance of invasive meningococcal disease (IMD) Neisseria meningitidis isolates and comparison with prevaccination isolates. Bexsero Antigen Sequence Types (BASTs) were used to analyze whole-genome sequences of 3,073 prevaccine IMD N. meningitidis isolates obtained during 2010−2016. Isolates exhibited 803 BASTs among 31 clonal complexes. Frequencies of antigen peptide variants were factor H binding protein 1, 13.4%; Neisserial heparin-binding antigen 2, 13.8%; Neisseria adhesin A 8, 0.8%; and Porin A-VR2:P1.4,10.9%. In 2015−16, serogroup B isolates showed the highest proportion (35.7%) of exact matches to >1 Bexsero components. Serogroup W isolates showed the highest proportion (93.9%) of putatively cross-reactive variants of Bexsero antigens. Results highlighted the likely role of cross-reactive antigens. BAST surveillance of meningococcal whole-genome sequence data is rapid, scalable, and portable and enables international comparisons of isolates.

In September 2015, 4CMenB meningococcal vaccine was introduced into the United Kingdom infant immunization program without phase 3 trial information. Understanding the effect of this program requires enhanced surveillance of invasive meningococcal disease (IMD) Neisseria meningitidis isolates and comparison with prevaccination isolates. Bexsero Antigen Sequence Types (BASTs) were used to analyze whole-genome sequences of 3,073 prevaccine IMD N. meningitidis isolates obtained during 2010−2016. Isolates exhibited 803 BASTs among 31 clonal complexes. Frequencies of antigen peptide variants were factor H binding protein 1, 13.4%; Neisserial heparinbinding antigen 2, 13.8%; Neisseria adhesin A 8, 0.8%; and Porin A-VR2:P1.4,10.9%. In 2015−16, serogroup B isolates showed the highest proportion (35.7%) of exact matches to >1 Bexsero components. Serogroup W isolates showed the highest proportion (93.9%) of putatively cross-reactive variants of Bexsero antigens. Results highlighted the likely role of cross-reactive antigens. BAST surveillance of meningococcal whole-genome sequence data is rapid, scalable, and portable and enables international comparisons of isolates. N eisseria meningitidis is an accidental human pathogen that is carried asymptomatically in the nasopharynx of 1%-40% of the population, depending on age and social behavior (1,2). In England and Wales, invasive meningococcal disease (IMD), comprising septicemia, meningitis, or both, develops in ≈2 persons/100,000 population/year (3). Patients might have nonspecific symptoms early in their illnesses, but their clinical conditions can deteriorate rapidly, with case-fatality rates of 5%-17% and physical and psychological sequelae in one third of surviviors (3)(4)(5). Consequently, vaccination represents the optimal strategy for IMD prevention.
Meningococci are commonly characterized according to their expressed capsular polysaccharides, which define serogroups; 6 serogroups (A, B, C, W, X, and Y) cause most cases of IMD. The capsular antigens are major virulence factors, and efficacious A, C, W, and Y polysaccharide−based vaccines are used worldwide. However, serogroup B capsular polysaccharides are poorly immunogenic and share structural similarity to carbohydrates found in human tissues (6). The first serogroup B vaccines were derived from outer membrane vesicles (OMVs) for use in epidemics caused by single strains defined by genotype and Porin A (PorA) type (7). Genotype or clonal complex (CC), identified by multilocus sequence typing (MLST), groups related organisms and is useful for categorizing IMD phenotype, antimicrobial drug resistance, and vaccine antigens (8,9).
The United Kingdom and Ireland are among high-income countries with the highest incidence of IMD (10). The United Kingdom has low-incidence endemic disease and periods of hyperendemicity, which changes with frequency of hyperinvasive bacterial genotypes (10). Historically, endemic serogroup B IMD predominated and was caused by multiple CCs, especially hyperinvasive lineages CC41/44, CC269, CC213, and CC32 (10). In the 1990s, hyperendemic serogroup C IMD caused by CC11 (C:CC11) prompted introduction of infant meningococcal C conjugate vaccination and a catch-up campaign, which reduced disease incidence and carriage of C:CC11 (11). The United Kingdom experienced another period of hyperendemicity starting in 2012 with increasing incidence of W:CC11, lineage 11.1 meningococci, first identified in South America (12).
A novel nomenclature, Bexsero Antigen Sequence Types (BASTs), was devised to describe Bexsero antigenic variants (22). There were strong, nonoverlapping associations between BAST and CC, with an estimated 58.3%-60.3% Bexsero coverage including the antigenic variants in Bexsero or cross-reactive variants (22). We cataloged genomic diversity of Bexsero vaccine antigens by using web-accessible platforms incorporating BAST. This study provides a reference point for changes in population structure of IMD-causing meningococci in the United Kingdom before introduction of Bexsero.

Materials and Methods
A total of 3,073 meningococci were isolated from culture-confirmed IMD cases in the United Kingdom during epidemiologic years (July 1−June 30) 2010-2016. For the purposes of this study, the prevaccine period includes 2015−16 because implementation of Bexsero started on September 1, 2015, and many infants were not fully vaccinated during the peak IMD season (December 2015−February 2016). The 3,073 isolates represented ≈55% of laboratory-confirmed cases of IMD (Table 1) because recovery of isolates reflects differential survival in artificial media, susceptibility to antimicrobial drugs given before venipuncture, or small-volume pediatric blood cultures.

WGS was part of the Meningitis Research Foundation
Meningococcus Genome Library initiative (10). Genomes were assembled by using Velvet and VelvetOptimiser, uploaded to the PubMLST database, and annotated by using Neisseria Sequence Typing Database numbers (NEIS) for all loci. Analysis was undertaken by using the gene-by-gene approach with the Bacterial Isolate Genome Sequence Database to determine sequence type (ST), CC, and strain designation (21,23). Each isolate had associated provenance and phenotype data, including year, serogroup, and region. We assigned BASTs as described (22). Nucleotide sequences of fhbp, nhba, nadA, and porA variable regions 1 and 2 (VR1 and VR2) were translated to deduce peptide sequences, and variant numbers were assigned by using established nomenclatures (24)(25)(26). Unique combinations of the 5 components were assigned a BAST number in order of discovery; BAST-1 corresponds to the vaccine constituents: fHbp 1, NHBA 2, NadA 8, and PorA 7-2,4 (22). Data were manually curated to confirm the absence of fhbp, nhba, nadA, and porA, and isolates were assigned peptide designation 0 (null). If nucleotide sequences contained a frameshift mutation, peptide designation 0 (null) was assigned. Peptide variants were not assigned if the complete gene was not available because of sequencing or assembly issues (22,27).
For assessment of Bexsero antigenic variants expected to be prevented by vaccination (coverage), we compared the genotypic profile (BAST) of isolates with those of vaccine antigenic variants. The term exact match indicates isolates having >1 of 4 Bexsero antigenic variants (fHbp 1, NHBA 2, NadA 8, and PorA-VR2:4). The term crossreactive match indicates isolates having >1 variant that can potentially be recognized by Bexsero-induced antibodies, demonstrating a possible cross-protective immune response in humans.
These variants were previously identified by using MATS analysis, which at the time of writing, was the most extensively used method for assessing qualitative and quantitive differences in antigens (20,28,29). Variants were considered putatively cross-reactive; fHbp peptides 4, 13, 14, 15, 37, 232 and NadA variants 1 or 2/3. NadA peptides were included because of potential discrepancies between in vitro and in vivo NadA expression (20,30). Cross-reactive NHBA peptides were not included because of lack of data on the breadth of peptides covered by Bexsero. Genomic analysis has not been used to infer protein expression or immunologic cross-reactivity per se.
We performed all statistical analyses by using R software version 3.2.4 (https://www.r-project.org/). We calculated the Simpson index of diversity by using the Vegan package in R software to assess diversity of BAST; values closer to 1 indicated greater diversity.

Nomenclature of Antigenic Variants
There are 3 systems for classifying fHbp variants. The first system described 3 variants (1-3) on the basis of sequence similarity and cross-reactivity in serum bactericidal antibody (SBA) assays (31). The second system described 2 subfamilies, A and B (32). The third system, used in our study as part of the BAST scheme, assigned arbitrary numerical integers to unique deduced peptide sequences independent of variant/subfamily (24). NHBA peptide variants were assigned arbitrarily to unique peptide sequences. Updated nomenclature for NadA described 4 variants on the basis of peptide sequence homology: NadA-1, NadA-2/3, NadA-4/5, and NadA-6 (26). PorA nomenclature was based on nucleotide and peptide sequence homology and recognized the previous serologic classification: P1, followed by VR1 family-variant, and VR2 family-variant (e.g., P1.7-2,4) (25). All nomenclature is available in the PubMLST database.

BAST Distribution by Serogroup
Although Bexsero is licensed for serogroup B IMD, its components might be present on the surface of meningococci, independent of capsular type. Therefore, we analyzed distribution of Bexsero antigens by serogroup. The antigenic variants were fHbp (50 isolates, predominantly peptide 13) and NadA (50 isolates, predominantly peptides 121, 127, and 1), largely reflecting secular changes in CC distribution with increases in CC11, CC269, CC32, and CC174.

Discussion
The requirement for vaccines to protect against serogroup B meningococci from multiple CCs led to development of multipeptide vaccines, such as 4CMenB (Bexsero) and bivalent rLP2086 (Trumenba) (16,17). Bexsero was introduced into the UK infant immunization schedule in September 2015, supported by data from the MATS assay that estimated 73% IMD N. meningitidis isolate coverage in England and Wales in 2007-08 (20).
We report a comprehensive evaluation of the frequency distribution of Bexsero antigen peptide variants in IMD N. meningitidis isolates, which used a national collection of WGS of culture-confirmed cases from 2010-2016. The frequency distribution of individual vaccine antigens was correlated with the distribution of meningococcal CCs in the United Kingdom over time. During 2010-2016, high diversity of 803 BASTs and 31 CCs emphasized the broad coverage required of peptide-based vaccines if they are to protect against endemic disease caused by multiple CCs. The most frequently occurring BASTs (20 representing 44.1% of serogroup B isolates) could provide useful information for future vaccine formulations.
Surface protein expression is also a major determinant of bactericidal killing. For fHbp, when heterologous bactericidal activity was tested, mouse antisera to peptide 1 produced positive titers against closely related peptide 4 regardless of expression level. For more distantly related peptides, such as peptide 15, higher protein expression was required (34).
At the time of writing, the most extensive estimation of cross-reactivity data for Bexsero had been collected by using the MATS assay. This assay quantifies expression and antigenic similarities of fHbp, NHBA, and NadA by sandwich ELISA and identifies PorA serosubtype by sequencing for each isolate (20,35). During development, the antigen measurements for fHbp, NHBA, and NadA were correlated to bactericidal killing with relative potency (RP) against 57 reference isolates tested by an SBA assay, determining likelihood of bacterial killing. Isolates with PorA-VR P1.4 peptide were considered to be covered, without further serologic testing (35). Contemporaneous MATS estimate of coverage for IMD N. meningitidis isolates from the United Kingdom during 2014-15 was 66% (95% CI 52%-80%) (28). However, the presence, cross-reactivity, and expression levels of antigenic variants in meningococci alone does not directly measure their susceptibility to bactericidal killing, which is also dependent on host innate and adaptive immune responses, a function not measured by the MATS assay. Therefore, this assay remains a surrogate for estimating functional activity against cross-reactive antigenic variants. Among UK isolates we examined, the most frequent peptide variant 1 fHbp peptides were 4, 13, 15, 14, and 1. For most fHbp peptide 1 isolates tested by MATS, their RP lies above the positive bactericidal threshold (PBT), and these isolates are predicted to be killed by the pooled serum from Bexsero-vaccinated toddlers used in the assay (20). However, for other fHbp variant 1 peptides, there was marked variation in coverage estimates by MATS for isolates with the same peptide variant. Two MATS studies in Europe identified the RP for peptide 4 isolates to be most consistently above the PBT, but RP for peptide 13, 14, and 15 isolates spanned the PBT (20,29). The degree of protection afforded by Bexsero vaccination will be observed through postimplementation enhanced surveillance. With 2-dose vaccine uptake at 88.6%, early reports of vaccine efficacy were estimated to be 82.9% (95% CI 24.1%-95.2%) (36). If these high efficacy estimates, albeit with wide CIs, continue to show protection beyond that predicted by genotypic, phenotypic, or functional estimates, then synergistic activity or minor antigens might need to be considered, neither of which are quantified by MATS or BAST.
Coverage for nonserogroup B isolates by Bexsero-induced immunity was of special interest in the United Kingdom because of increasing IMD cases caused by W:CC11 from 2012, with severe and atypical IMD and high mortality rates (37). The principal change in this analysis was the increase of BAST-2 (22; 29; 6; 5; 2), a direct consequence of W:CC11 clonal expansion. Although conjugate ACWY vaccine was introduced in August 2015, it was targeted to teenagers, the age group with increased disease and highest risk for carriage (38). There is a paucity of supporting immunologic evidence for the role of Bexsero in protection against nonserogroup B isolates, but in a small case series of 6 W:CC11 isolates, all BAST-2, human SBA assay responses of >1:32 were observed by using pooled serum from infants vaccinated with 3 doses of Bexsero (39). Therefore, some protection might be provided to Bexserovaccinated infants and toddlers.
After implementation of Bexsero into the UK immunization schedule, long-term vaccine effectiveness will be established by enhanced IMD surveillance accompanied by characterization of meningococcal isolates (36). The methods used here are rapid, standardized, open-source, and readily applied to different settings (22,40). Use of WGS in the Meningitis Research Foundation Meningococcus Genome Library initiative in the United Kingdom to extract vaccine antigenic variant data enables rapid isolate characterization, surveillance of circulating meningococci, and monitoring of secular changes and the impact of all meningococcal vaccines in use (10,22). These data serve as a reference point against which effects of the national Bexsero program can be compared, and highlight reliance on cross-reactive variants to maintain effective protection. Finally, such data will be invaluable in development of novel vaccine formulations that ensure continued coverage.