Invasive meningococcal disease in Shanghai, China from 1950 to 2016: implications for serogroup B vaccine implementation

Serogroup B invasive meningococcal disease (IMD) is increasing in China, but little is known about the causative meningococci. Here, IMD and carriage isolates in Shanghai characterised and the applicability of different vaccines assessed. Seven IMD epidemic periods have been observed in Shanghai since 1950, with 460 isolates collected including 169 from IMD and 291 from carriage. Analyses were divided according to the period of meningococcal polysaccharide vaccine (MPV) introduction: (i) pre-MPV-A, 1965–1980; (ii) post-MPV-A, 1981–2008; and (iii) post-MPV-A + C, 2009–2016. Over this period, IMD incidence decreased from 55.4/100,000 to 0.71 then to 0.02, corresponding to successive changes in meningococcal type from serogroup A ST-5 complex (MenA:cc5) to MenC:cc4821, and finally MenB:cc4821. MenB IMD became predominant (63.2%) in the post-MPV-A + C period, and 50% of cases were caused by cc4821, with the highest incidence in infants (0.45/100,000) and a case-fatality rate of 9.5%. IMD was positively correlated with population carriage rates. Using the Bexsero Antigen Sequence Type (BAST) system, fewer than 25% of MenB isolates in the post-MPV-A + C period contained exact or predicted cross reactive matches to the vaccines Bexsero, Trumenba, or an outer membrane vesicle (OMV)-based vaccine, NonaMen. A unique IMD epidemiology was seen in China, changing periodically from epidemic to hyperepidemic and low-level endemic disease. At the time of writing, MenB IMD dominated IMD in Shanghai, with isolates potentially beyond coverage with licenced OMV- and protein-based MenB vaccines.

Prevention of MenB IMD is challenging due to the poorly immunogenic serogroup B capsular polysaccharide and concerns about autoimmunity, due to its structural similarity to structures occurring on human tissue. To address this deficit, two protein-based vaccines, Bexsero (4CMenB) and Trumenba (bivalent rLP2086), have been developed and licensed in Europe and the USA 11,12 . Bexsero is composed of factor H binding protein (fHbp), Neisserial heparin-binding antigen (NHBA), Neisseria adhesin A (NadA), and outer membrane vesicles (OMVs) containing the porin PorA (meningococcal subtyping antigen) as a major immunogen, whereas Trumenba contains two fHbp-subfamily variants 13 . Both Bexsero and Trumenba may elicit protective responses across meningococci of diverse serogroup expressing these or related protein variants 14 . Two methods have been established to predict Bexsero coverage. (i) The Meningococcal Antigen Typing System (MATS) combines phenotypic and functional assays 12 ; however, it is time and labour intensive, requires toddler serum, and is only performed by specialist laboratories. (ii) The Bexsero Antigen Sequence Typing (BAST) which uses sequence data is a rapid, scalable, and portable genotypic approach, which catalogues deduced peptide sequences and matches to vaccine variants (BAST-1) or cross-reactive variants 15 . The meningococcal antigen surface expression (MEASURE) system is used to assess Trumenba coverage by flow cytometry 16 .
Limited information is available documenting N. meningitidis isolates associated with IMD and carriage in China over the past 60 years. In this study, fluctuations of IMD and meningococcal carriage are described in association with the introduction of MPV vaccines in Shanghai, China since 1950. In addition, and in response to increasing MenB IMD 7 , we assessed the potential impact of protein-based vaccines to local prevalent serogroups and clonal complexes.

Discussion
This study provides a comparative analysis of IMD and meningococcal carriage isolates obtained in Shanghai, China, and the likely impact of novel vaccines on disease and carriage. From 1965 to 1980, IMD in China was dominated by MenA meningococci belonging to the hyperinvasive lineages corresponding to cc5 (ST-5) and cc1 (ST-3), resulting in several epidemics (Figs 1 and 5), consistent with similar MenA epidemics seen elsewhere 17,18 . Introduction of serogroup A MPV vaccine in China in 1980 was followed by a decrease in IMD; however, this in turn may have contributed to expansion of MenC:cc4821 5 . Clonal expansion following vaccine implementation was further observed with the implementation of serogroup A and C MPV in 2008 which was followed by an increase in MenB IMD, largely due to MenB:cc4821 isolates (Figs 1 and 5, Table 2) 19 . These data indicate that  vaccine intervention may have promoted the emergence of different IMD-associated meningococci, consistent with the fluctuation of hyperinvasive lineages in the carried population; indeed, similar changes subsequent to vaccine implementation with MPV A + C were observed in Egypt and Morocco during 1992-1995 20 .
Since the 1950s, the seasonal peak of IMD cases in Shanghai has been from February to April (Fig. 3), similar that seen nationally in China 6 . China is a large country, with differences seen for example in the peak influenza season between northern (January) and southern China (from June to July), with the latter characterised by a warmer, more humid climate 21 . This suggests that, besides climate and influenza incidence, national social factors should be considered when deploying preventative strategies. IMD outbreaks, such as the MenA:cc5 global pandemic and MenW:cc11 transmission, are often associated with mass gathering events 17,22 . For example, the 1967 MenA epidemics across China (403/100,000) occurred following the National Great Networking event during 1966-1967 6,23 , which involved gatherings of millions of students from all over the country 23 . The observed seasonal peaks in IMD incidence (Fig. 3) may be associated with the Spring Festival travel rush for the Chinese New Year where, from January to March, over 200 million people travel across China for family reunions 24 . Poor sanitary conditions and overcrowded environments on public transport is likely to facilitate transmission of meningococci, indicating the need for enhanced understanding of meningococcal carriage before, during, and after the Chinese New Year.
Few carriage surveys have been undertaken in China; however, two studies in the Shandong and Guangxi provinces identified high carriage rates of meningococci from hyperinvasive ccs in association with IMD outbreaks 10,25 . This is consistent with results from our study where carriage rates positively correlated with incidence ( Fig. 4 and Table 1). In addition, MenB cases were not linked to a distinct seasonal pattern such that in the post-MPV-A + C period, MenB cases were found to occur throughout the year in contrast to that seen in the 1950s (Fig. 3).
Since the 1980s, three monovalent OMV-based MenB vaccines have been licensed for IMD epidemics, but these demonstrated clinical efficacy only against homologous meningococci 13 . Although a nonavalent OMV-based MenB vaccine has been evaluated 26 , the vaccine variants were found low prevalence among Chinese MenB IMD meningococcal isolates in this study (<5%) and from 27 provinces of China (<11%) 19 . Two protein-based MenB substitute vaccines have been licensed and implemented in vaccination interventions in Europe and the USA 11,12 . The coverage of MenB isolates by Bexsero in the UK during 2014/15 was predicted to be 60.8% using BAST 15 , and 66% by MATS 27 . For Trumenba, coverage rates of 78-100% to collections of diverse strains in Europe and the USA was estimated using serum bactericidal assay [28][29][30] . In this study, the prevalence of meningococcal antigens potentially covered by the variants present in the vaccine was low in Shanghai for both Bexsero (≤15%) and Trumenba (≤50%). Based on fHbp data from 30 provinces across China 31 , Trumenba was predicted to potentially cover 32.5% of IMD and 40% of MenB carriage isolates. This low prevalence is due to the predominance of cc4821 meningococci in China 19 , which share no homologous antigens with Bexsero and few with Trumenba (40.3%). In Europe and the USA, MenB IMD has been mainly caused by cc32, cc41/44, and cc269 over the past 20 years 32 , which exhibit distinct antigenic profiles to those found in China (Table 1). Therefore, the likely impact of Bexsero on Chinese cc32 (4.3%) and cc41/44 (42.9%) was lower than in Europe (93-100%) 27 . Alternative vaccination approaches include an OMV-based vaccine specific for MenB cc4821, and the characterisation of ST and antigen data, especially PorA variants, reported here will be invaluable in assessing vaccine coverage and future serogroup B-substitute vaccine development in China. The interpretation of these data is limited by incomplete records of MenB IMD cases during 1950-2004 and the small number of isolates collected during 1965-1980 and 1981-2008; 35 . Among these MenB cases in Shanghai and Beijing, 59.2% (84/142) occurred outside the "epidemic season" (from February to April). Therefore, it is likely the two features of MenB IMD, namely (i) the occurrence in young children and (ii) lack of seasonal variation in incidence, have persisted since the 1970s.
Our data suggest that vaccine coverage of MenB meningococci in Shanghai by licensed OMV and protein-based 'serogroup B substitute' MenB vaccines may be limited. Therefore, a cautious, region-specific approach to implementation of new protein-based meningococcal vaccines should be considered. Further, the temporal analysis suggests that vaccine implementation coinciding with the start of the calendar year, so as to disrupt transmission events during Spring Festival might have the highest impact on IMD incidence. In conclusion, our data indicate that IMD surveillance in China should be enhanced, combined with comprehensive carriage studies to assess the impact of vaccines and their likelihood inducing both effective direct protection and herd immunity.

IMD surveillance. IMD surveillance in Shanghai, implemented in the National Notifiable Diseases Registry
System (NNDRS), began in 1950 and was based on monthly paper reports. Since 2004, it has become a web-based, real-time system 6 . All clinical specimens and meningococcal isolates from suspected IMD cases in Shanghai are sent to Shanghai CDC when they are reported in the NNDRS 6 . In China, a child is defined as an individual <15 years of age and an infant <1 year 6 . N. meningitidis carriage surveys. Twenty carriage studies were conducted during 1965-2016. In each study, three districts were chosen, including urban, suburban, and rural districts. Posterior oropharyngeal swabs were collected from preschool children (toddlers aged 3-6 years in childcare centres), students (aged 6-14 years in schools), and adults (staff in department stores, railway stations, army, and residents in communities), and cultured as previously described 36 . Isolate collection. From 1965From -2016, 460 meningococcal isolates were collected in Shanghai, excluding the period 1986-2004 when isolates were not stored. As a result, 169 IMD and 291 carriage isolates dating from 1965-1985 (n = 306) and 2005-2016 (n = 154) were available for study. Serogroup was determined by slide agglutination using monoclonal antiserum (BD, USA) and PCR detection of the appropriate capsule synthesis gene 37 . Isolate serogroup distribution was: A, 123 isolates; B, 221; C, 62; E, 13; W, 5; X, 3; Y, 9; and Z, 3; and 21 nongroupable (NG, negative by PCR and sera agglutination). Sequence type (ST), cc, porA and fetA variants were determined from nucleotide sequences by querying the PubMLST.org/Neisseria database 38 . Relationships between STs were analysed using BioNumerics software package (version 7.6.2; Applied Maths, Belgium).
BAST identification and vaccine coverage estimates. BAST was determined as described previously 15 . Briefly, genome sequences were used to deduce peptide sequences by in silico translation of the nucleotide sequences of fhbp, nhba, nadA, and porA variable regions 1 and 2 (VR1 and VR2). Variant numbers were assigned to unique peptide sequences using the established nomenclatures [39][40][41] . Each unique combination of the 5 antigenic variants was assigned a BAST number, in order of discovery; BAST-1 refers to the Bexsero vaccine constituents: fHbp 1, NHBA 2, NadA 8, and PorA P1.7-2,4. Exact matches and potential cross-reactive matches were combined to evaluate coverage of Bexsero, Trumenba, and NonaMen, a 9-valent OMV-based vaccine (Table S2) 26,28-30,42,43 . Statistical analysis. Statistical analysis was performed using SPSS (version 20.0; IBM, USA). Fisher's exact test was used to compare proportions of IMD occurring in children with causative serogroup. Statistical significance was assessed at p < 0.05. The correlation coefficient between carriage rate and IMD incidence was calculated using Microsoft Excel 2010.
SCIentIfIC RepoRTS | (2018) 8:12334 | DOI:10.1038/s41598-018-30048-x Ethical Aspects. All the specimens from meningococcal patients and carriers were collected as part of the routine clinical management of invasive meningococcal cases, according to the national guidelines in China. Informed consent has been obtained, including that from a parent and/or legal guardian of the participants under the age of 18 years. All these experimental protocols were approved by the Institutional Review Board of Huashan Hospital, Fudan University. Data Availability. All data generated or analysed during this study are included in this published article and its Supplementary Information files.