Meningococcal disease is a notifiable disease in New Zealand; all IMD cases are referred to the Meningococcal Reference Laboratory at the Institute of Environmental Science and Research (ESR) for routine grouping using slide agglutination or PCR (21). For this study, disease incidence and case demographics were derived from notification data extracted from the national notifiable disease surveillance database. Annual population denominators were taken from Statistics New Zealand. We used R version 3.4.4 (https://www.r-project.org) to perform all statistical tests. Shannon–Wiener diversity index was calculated by using vegan version 2.5.6 (22).

We analyzed all available meningococcal isolates from 2013–2018 in New Zealand by WGS (Appendix 1 Table 1). Genomic DNA was purified by using the Gentra Puregene Yeast/Bact. Kit (QIAGEN, https://www.qiagen.com) or High Pure PCR Template Preparation Kit (Roche, https://www.roche.com) according to the manufacturer’s protocols. A 10-µL loop of bacteria, grown overnight on Columbia blood agar plates (Fort Richard Laboratories, Auckland, NZ), was suspended in 300 µL of lysis buffer and heat-killed at 56°C for 1 h. We then quantified DNA by using the Quant-iT PicoGreen dsDNA assay kit (Thermo Fisher Scientific, https://www.thermofisher.com) and constructed libraries by using Nextera-XT DNA Library Preparation Kit (Illumina, https://www.illumina.com). Paired-end sequencing of 2 × 150 bp was performed on the Illumina platform at ESR. Read data are available for download from the National Center for Biotechnology Information Sequence Read Archive (Bioproject accession no. PRJNA592848) and from PubMLST (https://pubmlst.org).

Raw reads were quality trimmed by using Trimmomatic version 0.32 (23) to remove adapters, low quality bases (<Q20), and reads shorter than 60 bp. SPAdes version 3.10.1 (24) was used for assembly, and contigs >200 bp were kept. Assembled contigs were used for in silico MLST, capsular grouping, and antigen typing by using meningotype version 0.82-β (25). Single-nucleotide polymorphisms (SNPs) were identified by aligning quality-trimmed paired-end reads to the N. meningitidis reference sequence FAM18 (GenBank accession no. AM421808.1) and using Bowtie2 version 2.2.5 (26). Alignments were processed with Picard-tools (27) to remove duplicated reads and assessed with Qualimap version 2.2.1 (28) (Appendix 1 Table 2). We used FreeBayes version 1.2.0–2-g29c4002 (E. Garrison, unpub. data, https://arxiv.org/abs/1207.3907) to detect sequence variations among isolates with these settings: ploidy = 1, at 20× minimum depth and 70% minimum variant allele frequency. We filtered variants by using vcflib (29) and vcftools version 0.1.12b (30) and removed sites located in the tandem repeat regions by using the intersect function from BEDTools version 2.23.0 (31). For datasets that included only assembled genomes, we used Parsnp version 1.1.2 (32) to perform core-genome alignment and FAM18 (AM421808.1) as reference.

We constructed phylogenetic analyses from core SNP alignment by using the maximum-likelihood method under the general time reversible substitution model with RAxML version 8.2.12 (33) and estimated the relative robustness of the clades with 200 bootstrap replicates (34). We used ClonalFrameML version 1.25 (35) to construct recombination-corrected phylogeny on the basis of core SNP tree topology. For datasets that included only assembled genomes, we generated a maximum-likelihood phylogenetic tree by using RAxML on the basis of the core-genome alignment with 200 bootstrap replicates. Phylogenies were annotated by using iTOL (36).

We compared New Zealand group W:CC11 isolates with 153 short-read datasets and 30 draft assemblies. Short-read data in the European Nucleotide Archive were selected to represent all major lineages of group W:CC11. We followed studies by Lucidarme et al. (14) and Tsang et al. (9) and randomly chose 1–3 isolates per year per country. We selected 153 datasets including 151 group W isolates and 2 group C isolates (ERR557598 and ERR976806), which were used as an outgroup to study the genetic relationship within group W:CC11.

We used the PubMLST Neisseria database to identify other isolates closely related to the New Zealand ST11 isolates (37). As of February 18, 2019, a total of 30 isolates were within the 50 mismatch thresholds to NZ17MI0022 based on comparison of the N. meningitidis core-genome MLST (38). Draft assemblies of these 30 isolates were downloaded from PubMLST. Short-read data were available for 3 of these 30 isolates. We compiled detailed information on sequences used for phylogenetic analysis (Appendix 1 Table 3).

Of the 45 New Zealand invasive group W:CC11 isolates, 42 were assessed for susceptibility to penicillin, ciprofloxacin, ceftriaxone, and rifampin. MICs were determined by gradient strip on Mueller-Hinton agar with 5% sheep blood and interpreted according to the Clinical and Laboratory Standards Institute breakpoints (39). Assembled contigs were used for in silico penA typing by using the PubMLST Neisseria database.

During 2013–2018, a total of 484 cases of meningococcal disease were reported, including 456 confirmed and 28 probable cases, for a confirmation rate of 94.2% (Figure 1, panel A; Appendix 1 Table 4). Rates of IMD (Figure 1, panel B) continued to fall after the end of the group B epidemic in 2008 (coefficient = −2.619, R² = 0.6947; p = 0.0245) but increased during 2014–2018 (coefficient = 2.481, R² = 0.9567; p = 0.0025). Inflection point of the time series was identified by breakpoint analysis and using the R package strucchange (40) and tested by the Chow test in R by using monthly disease rates (cases/100,000) during January 2008–December 2018 (Appendix 1 Table 5). The test identified the inflection point as October 2013 (95% CI January–December 2013).

Group B meningococci continue to be responsible for most IMD cases in New Zealand. Among the 438 confirmed cases analyzed by ESR during 2013–2018 were 265 (54.7%) group B (100 NZMenB and 165 other group B) cases, 58 (13.2%) group C cases, 61 (13.9%) group W cases, and 47 (10.7%) group Y cases (Figure 1, panel C; Appendix 1 Table 6). To compare whether significant changes occurred during 2013–2018, we used a 2-proportion Z-test in R to compare the proportion of diseases caused by each group. We found that the proportion of group W disease in 2018 (33/133) was significantly greater than that in 2013 (5/58, χ² = 8.2415; p = 0.002047). In contrast, the proportion of disease caused by group C in 2018 (10/133) is significantly less than that of 2013 (17/58, χ² = 10.578; p = 0.00057).

The rate of IMD continues to be highest among patients <1 year of age (from 10.2/100,000 population in 2014 to 23.1/100,000 population in 2017); rates are second highest among children 1–4 years of age (from 5.2/100,000 population in 2014 to 9.8/100,000 population in 2017). Adults >60 years of age are more affected by group W and Y disease (Figure 2; Appendix 1 Table 7).

We performed WGS and in silico typing on the 347 New Zealand meningococcal isolates (288 invasive and 59 noninvasive) collected during 2013–2018. The WGS dataset includes 175 group B isolates (65 NZMenB and 110 other group B), 48 group C isolates, 2 group E isolates, 64 group W isolates, 47 group Y isolates, 1 group X isolate, and 8 nongroupable isolates. MLST analysis showed that the 347 isolates contained 49 known STs. The 10 most common STs were ST11 (85), ST154 (43), ST23 (21), ST42 (17), ST32 (16), ST213 (15), ST1655 (15), ST1572 (11), ST6058 (8), and ST22 (7). There were 39 other STs with a single isolate and 19 unassigned STs (submitted through this project). The STs identified belonged to 18 clonal complexes. A total of 93 unique strains were defined by combination of group:PorA-variable region (VR)1, PorA-VR2:FetA:CC, with a Shannon–Wiener diversity index of 3.516. Bexsero antigen sequence types analysis showed 69 unique types with 23 isolates in which type could not be determined because of the absence of required loci (Appendix 1 Table 1).

Maximum-likelihood phylogeny constructed from 75,187 core SNP alignments (by using N. meningitidis FAM18 as the reference genome) showed 17 highly supported clades (with >90% bootstrap value), 15 of which were consistent with the CC definition (Figure 3). One clade of group B isolates could not be assigned to a CC because its ST had not been assigned a CC designation; it is noted as N1 on the phylogenetic tree. CC11 is the predominant CC consisting of 40 group C and 54 group W isolates; the second most common was CC41/44, consisting of 90 group B and 1 nongroupable isolates.

Since the NZMenB epidemic, New Zealand has used a combination of PorA subtype with group information to define N. meningitidis strains. We integrated the PorA and CC information to identify the strains currently circulating in New Zealand (Figure 4).

Rates of group W disease in 2018 were higher than in 2013. Most sequenced group W isolates in 2018 (30/35) belonged to CC11. To understand how New Zealand W:CC11 isolates relate to the global group W lineages, we analyzed New Zealand W:CC11 in the context of major W:CC11 lineages. We included 196 CC11 isolates in this analysis (151 downloaded from public databases and 45 invasive isolates from New Zealand). The mapping rate was 92.4%–99.3% when N. meningitidis FAM18 (AM421808.1), a representative of CC11, was used as a reference. The mean genome depth was 152X, with 84.8%–97.7% of loci covered at >20-fold (Appendix 1 Table 2).

We used core SNP (48,507 bps) alignment to construct the phylogenetic relationship of New Zealand W:CC11 isolates within the global W:CC11 (Figure 5). Phylogeny was rooted with 2 group C CC11 isolates. An unrooted neighbor-net phylogeny from the same dataset is shown in Appendix 2 Figure 1. Recombination-corrected phylogeny from the same dataset is shown in Appendix 2 Figure 2. Excluding the basal older sublineages, all other isolates formed 2 strongly supported clades. Clade I corresponds to the previously identified Hajj sublineage, and clade II corresponds to the previously identified South America sublineage. All New Zealand CC11 isolates were located within clade II: 4 clustered with the original UK strains, 12 clustered with the 2013 UK strain, and the other 29 formed a separate cluster with the 2 isolates from the United Kingdom and 1 isolate from Ireland (Figure 5).

To determine whether other isolates are closely related to the new New Zealand cluster, we searched PubMLST and found 27 additional isolates within 50 allele (the cgMLST 0.1 scheme) differences to NZ17MI0022. Because only assembled contigs were available for these isolates, we used a core-genome alignment approach to analyze their relationship within clade II of the W:CC11 isolates. All additional 27 W:CC11 isolates clustered with the New Zealand cluster with high bootstrap support (Appendix 2 Figure 3). The 27 isolates are from Canada and 6 countries in Europe (Appendix 1 Table 3).

Both phylogenetic analyses suggest that the new New Zealand cluster is part of the W:CC11 South America strain sublineage derived from the original UK strain. Because the earliest isolate of the new variant was identified in 2015, we named it the 2015 strain of the W:CC11 South America strain sublineage (the 2015 strain).

Similar to the 2013 UK strain, the 2015 strain is associated with higher death rates. During 2017–2018, the 2015 strain was associated with a death rate of 17.8% (6/34 cases) in New Zealand, higher than the rate of 5.9% (10/170 cases) for other groups in the same period (p = 0.03 by Fisher exact test). The 2015 strain also disproportionally affected older adults; 26% of total 2015-strain cases affected adults >60 years of age (9/34), whereas 5.8% (7/121) of total group B cases affected adults in that age group (p = 0.01466 by Fisher exact test).

In 2016, Mowlaboccus et al. (41) described a W:CC11 variant circulating in Australia that demonstrated intermediate resistance or was resistant to penicillin and had penA allele 253. To examine whether the New Zealand 2015-strain isolates were also resistant to penicillin, we tested the antimicrobial susceptibility of 42 invasive New Zealand W:CC11 isolates in this study. All 42 isolates were susceptible to ciprofloxacin (MIC <0.03 mg/L), ceftriaxone (<0.12 mg/L), and rifampin (<0.5 mg/L) (Appendix 1 Table 8). We observed variation in penicillin susceptibility among the 42 isolates (Appendix 1 Table 8) by using Clinical and Laboratory Standards Institute breakpoints. Of the 15 isolates belonging to either the original UK strain or the 2013 UK strain, all were susceptible to penicillin (<0.06 mg/L). Of the 27 isolates belonging to the new 2015 strain, 12 displayed intermediate resistance (0.12–0.25 mg/L) and 15 were resistant (>0.5 mg/L). The 2015-strain isolates were significantly more resistant to penicillin (p<0.00001 by Fisher exact test). All 27 New Zealand 2015-strain isolates had penA allele 253, the same allele described in the Australia study (41). In the larger dataset that included 30 international 2015-strain isolates, all but 1 isolate had penA allele 253 (Appendix 1 Table 3).