Surveillance of invasive Neisseria meningitidis with a serogroup Y update , Sweden 2010 to 2012

An increase of invasive meningococcal disease caused by Neisseria meningitidis serogroup Y has been noted in Sweden since 2005, and to a lower extent throughout Europe. The present study describes the epidemiology of invasive N. meningitidis isolates in Sweden in the period between 2010 and 2012, with a focus on serogroup Y. We also aimed to find an optimal molecular typing scheme for both surveillance and outbreak investigations. All invasive N. meningitidis isolates in Sweden during the study period (n=208) were genetically characterised. Serogroup Y predominated with 22/57, 31/61 and 44/90 of all invasive isolates (incidence 0.23, 0.33 and 0.46 per 100,000 population) in 2010, 2011 and 2012 respectively. In each of these years, 15/22, 22/31 and 19/44 of serogroup Y isolates were genetically clonal (Y: P1.5–2,10–1,36–2: F4–1: ST-23(cc23), ‘porB allele 3–36, fHbp allele 25 and penA allele 22). Our findings further support those of others that currently recommended FetA typing could be replaced by FHbp. Moreover, in line with a previous study that we conducted, the current results indicate that highly variable multilocus variable-number tandem repeat analysis (HV-MLVA) can be used as a firsthand rapid method for small outbreak investigations.


Introduction
Neisseria meningitidis (the meningococcus) is a Gramnegative diplococcus carried asymptomatically in the pharynx by approximately 10% of the population [1].It is also a potentially devastating pathogen causing meningitis and septicaemia.Invasive meningococcal disease (IMD) occurs mainly in sporadic cases but also as outbreaks and epidemics.Meningococcal populations are genetically and antigenically highly diverse [2] and vary greatly globally and over time, but the majority of IMD is caused by a limited number of clonal complexes, known as hyper-virulent lineages [3].Therefore, detailed characterisation of circulating meningococcal strains is important in terms of vaccination policy decisions, outbreak management, as well as monitoring antibiotic susceptibility and vaccine coverage.
The polysaccharide capsule surrounding the bacterium defines the meningococcal serogroup.The capsule is an important virulence factor and IMD is mainly restricted to encapsulated meningococci belonging to serogroups A, B, C, W, X and Y.The capsule is also a polysaccharide vaccine component in available conjugate vaccines for serogroups A, C, W and Y [4].The serogroup distribution is highly regional [5].In Europe, the main circulating strains belong to serogroups B and C [6].As previously described, serogroup Y has increased in Sweden from 0.04 per 100,000 population in 2005 to 0.23 per 100,000 population in 2010 [7].An emergence of serogroup Y has also been noted in some other European countries, however, the highest relative proportions are found in Scandinavia [8,9].
In addition to serogroup designation, it is currently recommended by the European Meningococcal Disease Society (EMGM) that meningococcal strains are designated by variable regions (VR) in the Porin A (PorA) and the Ferric enterobactin transport protein A (FetA) proteins as well as multilocus sequence typing (MLST) sequence type (ST) and clonal complex (CC) [10].PorA and FetA are two surface antigens, which are recommended for rapid investigation of disease outbreaks.MLST, based on seven housekeeping genes, is ideal for studying population biology and evolution of the organism on a national and international level.For enhanced resolution, genotyping of a third surface antigen, Porin B (PorB), may also be performed [11].Finally, further characterisation can be achieved with the penA gene encoding the penicillin-binding protein 2 (used in surveillance of penicillin susceptibility) [12] and fHbp encoding the serogroup B vaccine component Factor H binding protein (FHbp) [13,14].
Another molecular method that has been proposed for an alternative typing paradigm, mainly suited for investigating localised outbreaks, is multilocus variable-number tandem repeat analysis (MLVA) [15][16][17][18].MLVA is a polymerase chain reaction (PCR)-based technique, which uses the variability in the numbers of short tandem repeats to create DNA fingerprints used in epidemiological studies.The highly variable MLVA (HV-MLVA) developed by Schouls et al. [18] has shown high discriminatory capacity for serogroup C isolates and has been considered suitable for outbreak identification [19].
The aims of the present study were to describe the current epidemiology of invasive N. meningitidis isolates including the dominating serogroup Y in Sweden, and to find an optimal molecular typing scheme with appropriate resolution power for both surveillance and outbreak investigations.

Bacterial isolates and phenotypic characterisation
In Sweden, all invasive cases of meningococcal disease according to the European Union case definition are mandatorily reported by clinicians to the Swedish Institute for Infectious Disease Control (SMI) [20].The corresponding isolates are sent to the Public Health Agency of Sweden, where they are routinely cultured on chocolate agar at 37°C with 5% CO 2 overnight and subsequently serogrouped by co-agglutination [21].Further genosubtyping (PorA typing) is then conducted as previously described [22] and antibiotic susceptibility determined using the Epsilometer (E)test method (bioMérieux, Marcy l'Etoile, France).Basic epidemiological data (age, sex, area of residence, clinical site of isolation and date of sample collection) are gathered for all isolates from cases.This study included all invasive N. meningitidis isolates in Sweden between 2010 and 2012.The serogroup B strain MC58 [23] was included in all analyses as a reference.

Nucleic acid extraction
The DNA used for sequence-based typing methods and MLVA was extracted with a NorDiag Bullet instrument (DiaSorin, Dublin, Ireland).For the automatic extraction, 20 colonies from each cultured organism were suspended in 2 ml NaCl (0.85%) and 100 µl of this solution were subsequently processed with the NorDiag Bullet with the Bullet BUGS'n BEADS kit according to the manufacturer's recommendation (DiaSorin).All DNA preparations were stored at 4°C prior to the PCR.

Polymerase chain reaction and DNA sequencing
The MLST genes: abcZ, adk, aroE, fumC, gdh, pdhC and pgm together with fetA, fHbp, porB and penA were amplified and sequenced as previously described [7,19].In short, the PCR was performed using a Rotor-Gene Q real-time PCR system (Qiagen, Hilden, Germany) and detected with SYBR green I.The nucleotide sequences were determined by capillary electrophoresis using an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) using BigDye Terminator v3.1 (Applied Biosystems, Warrington, UK).The sequence alignments for the MLST genes were assembled using the Bionumerics software version 6.6 (Applied Maths, Sint-Martens-Latem, Belgium, http:// www.applied-maths.com/bionumerics)and all other genes were assembled using the ChromasPro software version 1.33 (Technelysium Pty Ltd, Brisbane, Australia, http://technelysium.com.au/).The different sequences were assigned allele numbers using the N. meningitidis sequence query database [24].

Multilocus variable-number tandem repeat analysis
The HV-MLVA with four highly variable variable-number tandem repeat (VNTR) loci (VNTR4-4, VNTR9-2, VNTR4-2, and VNTR4-3) was performed as previously described by Törös et al. [19] with the primers from Schouls et al. [18].In short, the PCR was performed on an Applied Biosystems 9700 or 2720 PCR machine and the fragments were separated on an ABI PRISM 3130xl Genetic Analyzer using the GeneScan LIZ 500 size standard and GeneScan module with filter set G5 (Applied Biosystems).The sizing of the fragments was performed with the GeneMapper software v4.0 (Applied Biosystems).All isolates were run in duplicates in separate runs.

Data analysis
The ability of each method to discriminate between strains was evaluated on the basis of their discrimination index, using Simpson's index of diversity (ID) [25].The ID determines the probability that two randomly picked strains are allocated to different types.Confidence intervals (CI) of 95% were calculated [26].
Cluster analysis of the MLST data was performed using a categorical coefficient and displayed in a minimum spanning tree (MST) created with the Bionumerics software version 7.1, with the priority rule of first linking types which have the highest number of single-locus variants.For the HV-MLVA, a dendrogram was generated using a categorical coefficient and unweighted pair group method with arithmetic average (UPGMA).

Resolving power of molecular typing methods used for surveillance
The discriminative ability of a MLST and the porA, porB, fetA, fHbp and penA genes shows that as single targets PorA VR 1, 2, and 3 had the highest Simpson's ID (0.849; 95% CI: 0.811-0.886)and serogroup had the lowest (0.664; 95% CI: 0.628-0.700]).Among combinations of four, on the basis of serogroup, porA and MLST, this combination including fetA had the highest

Figure 4
Minimum spanning tree from multilocus sequence typing profile data for Neisseria meningitidis isolates in Sweden, 2010-2012 (n=208) Each circle represents a sequence type (ST), and the serogroup distribution in each ST is represented by the respective colour.Lines connecting the STs that are thick and solid, thin and solid, dashed and dotted denote 1, 2, 3 and 4 or more loci differences, respectively.Halos surrounding the circles denote different clonal complexes.Purple: clonal complex ST-11; yellow: clonal complex ST-41/44; green: clonal complex ST-22; blue: clonal complex ST-213; peach: clonal complex ST-23; pink: clonal complex ST-32; turquoise: clonal complex ST-269.

Outbreak investigations
A spatiotemporal association was defined as isolates of the same serogroup, collected in the same central county within a timeframe of one month.Of the 208 N. meningitidis invasive isolates from 2010 to 2012, 35 isolates were spatiotemporally associated in 16 different clusters and represented potential outbreaks.There were 13 clusters of two isolates and three clusters of three isolates.Eight of the spatiotemporal clusters (cluster nr 1-8 in Figure 6) did not share a common PorA type within the clusters and the cases had not been in direct contact with each other and therefore there were no further outbreak investigations.In spatiotemporal clusters nr 10-13, the isolates were identical within each cluster regarding serogroup and all 12 target genes (Figure 6), but no connection between cases was found and the isolates were separated by HV-MLVA.Within each of the three spatiotemporal serogroup B clusters 14-16 (Figure 6) the isolates were identical regarding all 12 sequenced genes (the isolates in cluster 14 all had one insertion in aroE allele 9 but were still regarded as belonging to ST-41).In addition, the isolates were clustered together in the HV-MLVA when MLVA types did not differ in more than one VNTR locus (single-locus variant, SLV).However, spatiotemporal cluster 14 and 16 were the only spatiotemporal clusters which had confirmed connections between cases.The HV-MLVA results from all 208 isolates showed another seven HV-MLVA clusters comprising fifteen isolates in total (if SLVs were allowed), of which four did have identical genetic profiles, but none of them shared a spatiotemporal link and were therefore not included in Figure 6.

Discussion
This study aimed to describe the epidemiology of invasive N. meningitidis isolates in Sweden between 2010 and 2012, specifically the dominating serogroup Y, and to identify an optimal molecular typing scheme with appropriate resolution power for both surveillance and outbreak investigations in low-endemic areas.
Although whole genome sequencing is becoming more cost beneficial than traditional Sanger sequencing per isolate, optimal molecular typing schemes will still be important when fast results are needed, and for smaller laboratories that lack the need of a next generation sequencer.All isolates causing IMD in Sweden during 2010, 2011 and 2012 were characterised by capsular group, MLST, sequencing of the porA, fetA, porB, penA, fHbp genes and a MLVA using four highly variable loci.
The epidemiology in Sweden has changed most notably with an increase of IMD between 2010 and 2012.IMD caused by serogroup C has declined slightly from 2010 to 2011 and 2012, and serogroup B somewhat increased (Figure 1).The genetic characterisation of circulating N. meningitidis causing IMD shows that serogroup Y was the most prevalent in Sweden, and the previously genetically described predominant clone YI [7] was still dominating among serogroup Y strains.However, the overall IMD incidence is still low and although a vaccine against serogroup Y is available, it is probably not FetA: Ferric enterobactin transport protein A; PorA: Porin A; SG: serogroup; ST: sequence type; VR: variable region.A similarity line has been drawn at 75% (maximum one single-locus variant).Only spatiotemporally associated isolates (Sweden, 2010(Sweden, -2012) ) are included and have been designated a spatiotemporal cluster number 1-16.Additional information about serogroup and allele or VR from sequencing the porA, fetA, 'porB (partial fragment), fHbp and penA genes and the ST from the multilocus sequence typing is provided for each strain.The resolving power of a molecular typing method is recommended to have a discrimination level of at least 0.90 [25].Serogroup:porA:fetA typing has previously shown a value of 0.963 [29] which is similar to the discrimination index achieved in this study (0.952; 95% CI: 0.935-0.969)with the same typing targets (data not shown).Although serogroup:porA:fetA:ST had the highest index of diversity achieved in this study, our results show that replacing the fetA gene with fHbp only reduces the discrimination ability by 0.006 (no statistical significant difference).Concurrently, Lucidarme et al. [30] investigated the comparability between fetA and fHbp in terms of diversity, after it had been recommended in England and Wales to additionally incorporate fHbp for routine genotypic surveillance.
Their study (on 613 invasive isolates) actually showed that fHbp had significantly (non-overlapping 95% CIs) better resolving power than fetA.These findings indicate that the level of discrimination gained from fHbp is partially complementary to that of fetA.This could strengthen the argument that considering labour and cost, and with regard of the new serogroup B vaccines, it would be beneficial to substitute the current routine marker fetA with fHbp.
In terms of outbreak investigations, HV-MLVA detected three small clusters with spatiotemporal connections and identical genetic profiles (spatiotemporal clusters 14-16 in Figure 6), which further supports that these strains were truly involved in small outbreaks.
Although no connection could be confirmed between cases in spatiotemporal cluster 15, the cases were both of similar age and from the same county and thus the cases could still have been related.
Fifteen isolates in the present study were clustered in seven HV-MLVA clusters without having a spatiotemporal connection (data not shown).However, without a spatiotemporal connection, these would normally never have been subjected to a HV-MLVA analysis.Moreover, the results of the HV-MLVA suggest that, after identifying the capsular group and receiving clinical data, it could have been sufficient to perform only the HV-MLVA to get indications of potential outbreaks.In an outbreak situation where time is of high importance, not having to await the results of PorA typing or whole genome sequencing for the subsequent HV-MLVA analysis would be beneficial.
We have previously compared HV-MLVA to rep-PCR with the DiversiLab system and DNA sequencing on invasive serogroup C isolates which showed that HV-MLVA helped strengthen all of the spatiotemporal linkages [19].The use of MLVA to trace transmission has been deemed questionable due to the low stability of VNTRs.Transmission-dependent variation of tandem repeats in meningococci has been investigated by Elias et al. [31] in a study using four highly variable VNTR loci together with another eight standard MLVA loci.The observed overall variation was considerably smaller than predicted and the method was considered most useful for outbreaks containing few transmissions.Considering this, HV-MLVA could be valuable in low-endemic areas such as Sweden where outbreaks are fairly rare and connected cases usually only consists of no more than three contacts, as shown by our results.A fast detailed typing of spatiotemporally linked cases which can separate outbreaks from sporadic cases is important to inform public health measures to control IMD, such as deciding whether or not to offer prophylaxis in the form of antibiotics or vaccines.
In summary, serogroup Y was found to be the most prevalent serogroup in Sweden, and the previously genetically described predominant clone YI [7] (Y: P1.5-2,10-1,36-2: F4-1: ST-23(cc23), 'porB allele 3-36, fHbp allele 25 and penA allele 22) is still dominating.However, a sulfamethoxazole susceptibility subvariant of the clone appears to have emerged in 2010, and in 2011 represented 11/22 (50%) of the genetically defined predominant clone YI isolates.Our study supports previous studies that suggests that the FetA typing could be replaced by FHbp in the recommended designation (serogroup:porA:fetA:ST(CC)), which may be more suitable in the current vaccine era.Furthermore, this study including the additional serogroups A, B, E, W and Y, strengthens previous results on serogroup C isolates [19] suggesting that HV-MLVA is a first-hand rapid method for investigating outbreaks with few transmission events, where isolates have an identical serogroup and a common spatiotemporal connection.

Figure 1 Figure 2
Figure 1 Incidence of invasive meningococcal disease caused by Neisseria meningitidis serogroups B, C, W and Y in Sweden, 2000-2012 (n=642)