INTRODUCTION
Haemophilus influenzae is an important cause of both adult and child respiratory disease and severe invasive disease. Strains are characterised largely by the presence or absence of a polysaccharide capsule (CPS). Encapsulated strains (Hi) are further classified serologically, possessing one of six different capsular types, designated a to f. Prior to the inclusion of the H. influenzae b (Hib) vaccine in the Australian National Immunisation Program Schedule in 1993, Hib was a common cause of invasive infections, particularly in children. Dramatic reductions in the rates of Hib disease has been noted in countries which have incorporated this vaccine into their vaccination schedules [Reference Falla1, Reference Jordens and Slack2].
Un-encapsulated H. influenzae, also termed non-typeable (NTHi) do not possess the CPS and as such cannot be classified by serotyping. The CPS is an important virulence factor of encapsulated strains [Reference Gilsdorf3], however a number of adhesive factors, IgA-proteases and cell surface glycolipids have been identified in NTHi that promote colonisation and influence host cell interaction and disease [Reference O'Neill4, Reference De Chiara5]. The possession of such genes can contribute to enhanced virulence in some strains. Many recent studies in particular have observed an increase in the frequency of invasive NTHi cases worldwide [Reference Bajanca-Lavado6–Reference Wan Sai Cheong12].
Hib is a notifiable disease in Queensland (QLD) and the majority of laboratory confirmed invasive Hi isolated by hospital and private pathology laboratories in the state are forwarded to the QLD Public Health Microbiology (PHM) reference laboratory for serotyping. A large proportion of referred invasive QLD H. influenzae isolates are un-encapsulated however there is a distinct lack of published data on the genetic basis and population structure of both QLD and Australian NTHi strains. The aim of this study was to molecularly characterise referred invasive NTHi isolates in QLD, Australia over a 15-year period, observing the carriage of capsular and putative virulence genes including the major adhesins, antimicrobial resistance genes and population diversity.
METHODS
Specimen selection
Between 2001 and 2015, 745 invasive H. influenzae isolates from QLD patients (excluding duplicates) were referred to the QLD PHM laboratory. Invasive disease was defined as the isolation of H. influenzae from a normally sterile body site in a patient such as blood or cerebrospinal fluid (CSF). As part of reference laboratory surveillance, demographic information including age, gender and postcode was collected, as well as antimicrobial resistance data as recorded by the submitting laboratory using CLSI (Clinical and Laboratory Standards Institute) guidelines [13] for most isolates. Clinical presentation and disease outcome was not routinely provided to the reference laboratory. Isolate identity was confirmed using the RapID™ NH system (Remel, Kansas, USA). Isolates were classified into serotypes a–f (Hia–Hif) or as NTHi using commercially available sera (Phadebact Haemophilus Test; MKL Diagnostics AB, Sollentuna, Sweden and Denka Seiken; Tokyo, Japan). β-lactamase production was determined by the nitrocefin biochemical test (Oxoid, Hampshire, UK).
Further characterisation was performed on a subset of 100 isolates chosen to represent varying serotype, year of isolation, patient age and geographical location based on the total isolates received. NTHi (n = 74) formed the majority of isolates with the remaining 26 isolates representing serotypes a (n = 5), b (n = 5), c (n = 4), d (n = 2), e (n = 5) and f (n = 5).
Additionally, 11 publicly available complete genome sequences from encapsulated and un-encapsulated isolates were downloaded from GenBank in June 2016 for use in this study ((Name; GenBank accession no.): Rd_KW20; NC000907, KR494; CP005967, 86-028NP; NC_007146, PittEE; NC_009566, F3031; NC_014920, F3047; NC_014922, 10810; NC_016809, R2866; NC_017451, R2846; NC_017452, 477; NZ_CP007470 and C486; NZ_CP007471).
Whole-genome sequencing of isolates
Genomic DNA was extracted from the isolates using QIAsymphony AP (Qiagen, København, Denmark) as per the manufacturer's instructions. Genomic DNA was sheared to 300 bp using the Covaris S220 ultrasonicator and libraries were prepared using the Ion Plus Fragment Library kit. Sequencing was performed using an Ion Torrent PGM with the Ion PGM IC 200 kit using 316v2 chips (Life Technologies, Carlsbad, USA) according to manufacturer's instructions. FASTQ reads are located in the European Nucleotide Archive in study PRJEB18702.
Interrogation of assembled sequences for capsular and putative virulence gene detection
Sequences were de novo assembled using the Geneious assembler with default settings in Geneious® version 7.1.8 (Biomatters, Auckland, New Zealand) [Reference Kearse14]. Assemblies were imported into SeqSphere+ version 3.2.1 (Ridom GmbH, Münster, Germany) and task templates were developed for the detection of targeted putative virulence genes including major proteases and adhesins (iga, igaB, hmw1A, hmw2A, sodC, hap, hia and pepN), capsule-associated genes (bexA, bexB, capA, capB, capC, capD, capE and capF), and mobile element IS1016. For gene detection, the CDS (complete coding sequences) were downloaded from GenBank using the following reference isolates; 10810 (bexA, bexB), Rd_KW20 (iga, hap, pepN), R2866 (hia), R2846 (hmw1A, hmw2A), GenBank accession no. AF549211 (sodC), GenBank accession no. NC_022356 (IS1016 and capF), GenBank accession no. Z37516 (capA), GenBank accession no. FQ312006 (capB), GenBank accession no. HQ651151 (capC), GenBank accession no. HQ424464 (capD), GenBank accession no. HM053635 (capE). Detection of the hif gene cluster was achieved using a primer sequence for the conserved hifBC region from R2866 [Reference Ecevit15, Reference Mccrea16].
Targets were recorded by SeqSphere+ as present if found in the genome within the parameters of ⩾95% sequence identity and ⩾99% alignment to the reference sequences. For the purposes of this study, only gene detection or non-detection information was recorded and further investigation into the gene functionality of variants was not performed. The sequences for targets iga, hap, hia, hmw1A and hmw2A were shown to be highly variable and gave consistently low nucleotide sequence similarity to the references used (<95% identity and <99% alignment) and were recorded as absent by SeqSphere+ and flagged as possible false-negative results. In order to confirm the presence of these targets the reads of all isolates were mapped to the publicly available reference sequences for PittEE, Rd_KW20, R2866 and R2846 using CLC Genomics workbench 8 (Qiagen) which comprised representative nucleotide sequences of the investigated target genes respectively.
Multi locus sequence typing (MLST)
The MLST was determined for each isolate, facilitated using SeqSphere+, which assigned alleles according to the scheme at PubMLST.org. Data were imported into BioNumerics 6.5 software (Applied Maths Inc., Sint-Martens-Latem, Belgium) and a minimum spanning tree (MST) was produced using a categorical coefficient.
Single nucleotide polymorphism (SNP) analysis of isolates
FASTQ files for all isolates from this study, 10 GenBank Haemophilus influenzae genome sequences (KR494, 86-028NP, PittEE, F3031, F3047, 10810, R2866, R2846, 477 and C486) and the reference genome, Rd_KW20 were used to produce an alignment of core SNPs using SNIPPY version 3 [Reference Seemann17]. A maximum likelihood tree was generated from the SNP alignment, using the Fast Tree and visualised using the Phandango web application (https://github.com/jameshadfield/phandango).
Identification of antimicrobial resistance genes
The presence of acquired resistance genes was investigated for all isolates using the ResFinder tool at CGE (https://cge.cbs.dtu.dk/services/ResFinder-2.1/).
Data analysis
Diversity was calculated using Simpson's index of diversity (SID) [Reference Simpson18, Reference Hunter and Gaston19], with 95% confidence intervals (CI) calculated as previously described [Reference Grundmann, Hori and Tanner20]. A BioNumerics script available at http://biomath.itqb.unl.pt/ClusterComp [Reference Carriço21] was used to facilitate these calculations. Calculation of Yates' χ2 statistic was calculated as per Preacher [Reference Preacher22].
Postcode data were categorised according to QLD Hospital and Health Services (HHS) and was used to assign isolates to one of three regions; North QLD (Torres and Cape, Cairns and Hinterland, North West and Townsville HHS), Central QLD (Central West, Central QLD, Wide Bay and Sunshine Coast HHS) and South East QLD (South West, Darling Downs, West Moreton, Metro North, Metro South and Gold Coast HHS).
RESULTS
Invasive H. influenzae isolates referred
A detailed breakdown of data on isolates referred is shown in Table 1. The majority of the 745 invasive Hi isolates referred to the PHM laboratory between 2001 and 2015 were NTHi (72·2%). Of the 207 capsulated Hi, Hib were most common (11·1%), followed by Hif (6·9%), Hia (4·8%) and Hie (4·2%). Of the total referred isolates, 53·4% (n = 398) were from female patients and 46·2% (n = 344) from males, with three isolates having no gender recorded.
Table 1. Data breakdown for Haemophilus influenzae referrals to PHM laboratory 2001–2015
The nitrocefin test for β-lactamase presence was positive for 20·8% of total isolates, a similar rate was observed for NTHi isolates alone (21%). Interestingly, different serotypes exhibited different levels of β-lactamase activity; most notably Hia, Hic and Hid had no β-lactamase-positive isolates, while 30% of Hib were β-lactamase positive.
Location data by postcode were unavailable for 39 isolates. Of the remaining 706 isolates 56·6% were collected from South East QLD, 21·8% from North QLD and 21·6% from Central QLD. When differentiated by serotype, NTHi, Hib, Hie and Hif were predominately from South East QLD, however Hic and Hid were exclusively isolated from North QLD patients, and 65·7% of Hia were also isolated in North QLD.
Overall, patient age for QLD invasive Hi cases ranged from 0 to 99 years of age (median age 42·9 years). In total 23·4% of referred isolates belonged to patients aged ⩽5 years with 9·8% of these being from patients ⩽1year of age. Patients in the 61–80 year age group made up the second largest percentage of referred Hi (23·1%). When differentiated by type, Hie and Hif demonstrated similar age breakdowns. Interestingly NTHi were less frequently isolated in children ⩽5 years (17·7%) and had highest frequency from patients 61–80 years of age (25·1%). Other encapsulated serotypes were isolated more frequently from the ⩽5 years demographic (Hia 63·9%; Hib 37·4%; and Hic 75%).
An increasing trend was observed for the absolute number of NTHi isolates referred each year, with encapsulated strain numbers remaining fairly constant over the time period. A 13% year-on-year increase was observed for NTHi referral numbers from 2001 to 2014, with a 65·2% increase from 2014 to 2015, however this increase in NTHi as a proportion of all referred invasive Hi from 2001 to 2015 was not statistically significant (63% to 83·5%; χ2 Yates = 0·42; P = 0·52).
Detection of capsule-related genes in sequenced isolates
All encapsulated isolates possessed the expected capsule gene for their known serotype, along with the bexA and bexB genes, and the insertion sequence IS1016 (absent in one Hid isolate). The bexA and/or bexB were not detected in any NTHi isolates, although one NTHi isolate (Hi26) possessed the insertion sequence IS1016. Further investigation by ordering and aligning the de novo assembled contigs for Hi26 to the Hia reference sequence (GenBank CP017811) using Mauve [Reference Rissman23] demonstrated that this isolate closely matched the overall structure and content of the Hia reference, but did not contain the capsular region. IS1016 was present in Hi26 at the location of the deleted capsule locus and so is likely to be associated with the deletion of this capsular region.
Detection of putative virulence genes
Presence/absence reporting of putative virulence genes identified 10 separate gene profiles among the 100 isolates (SID = 0·81; 95% CI 0·77–0·85, data not shown). Table 2 shows the proportion of genes detected across each serotype. Genes iga, hap and pepN were ubiquitous among all isolates, while all other genes were present in a subset of the isolates. All encapsulated isolates lacked hmw genes and possessed hia. Encapsulated isolates largely clustered together with similar gene profiles, with Hie and Hif grouping exclusively together in one gene profile that also possessed sodC.
Table 2. Putative virulence genes possessed by serotype
The igaB was detected only in a proportion of NTHi isolates, which all formed one gene profile also having both hmw genes. One NTHi isolate possessed sodC in addition to both the hmw genes. The hia was detected in 65% (13/20) of hmw-negative isolates. Only seven isolates lacked hmw and hia, one of these carried hifBC, the other six possessed only iga, hap and pepN.
MLST of sequenced isolates
MLST identified 62 unique sequence types in total among the 100 isolates (SID = 0·99; 95% CI 0·98–0·99). Overall, eight novel alleles were identified (frdB, 176–177; fucK, 120; mdh, 266; pgi, 233, 244–245; recA, 146) and 18 novel STs (1462, 1463, 1472, 1508–1511, 1589–1599) were assigned.
Encapsulated isolates clustered together and were largely separate from NTHi isolates (Fig. 1). Eight clonal complexes (CCs) were identified, with isolates within each CC differing only by one locus. Five CCs (1–5) comprised NTHi isolates exclusively with the largest CC containing 10 isolates, two CCs were comprised only Hib (CC6) and Hie (CC7) respectively and the final cluster (CC8) contained all five Hia isolates and one NTHi (Hi26). No notable association with patient age, gender or location was demonstrated by cluster analysis.
Fig. 1. Minimum spanning tree based on allelic profiles obtained by MLST using a categorical coefficient. The 11 publicly available reference sequences have also been included (labelled in blue). Additionally, serotype data of significant clusters and encapsulated isolates are labelled. Each node represents a different ST, with the size of nodes proportional to the number of included isolates. Thick dark lines separate types differing by a single locus, while thin continuous lines connect double-locus variants. Faint lines represent types differing by three or more loci. Shading around nodes demonstrates partitioning of isolates differing by only one locus, and indicates designated CCs (1–8).
The largest single ST for QLD invasive NTHi was ST103 (n = 6), a further 50 STs were associated with the 67 remaining NTHi isolates, which were dispersed widely across the MST. Only one NTHi (Hi26) shared a ST with Hi isolates (ST56). With this exception, Hi fell into exclusive STs: Hia; STs 56 and 1511, Hib; STs 6 and 118, Hic; STs 19, 1508 and 1598, Hid; ST10, Hie; ST18 and 66 and Hif; STs 124 and 16.
SNP typing
The classification of isolates by SNP groupings correlated well with serotype, MLST ST and the distribution of putative virulence genes (Fig. 2). No notable correlation was observed between SNP groupings and year, age, location or gender (data not shown).
Fig. 2. Genetic relatedness, characteristics and putative virulence gene profile of invasive QLD H. influenzae isolates. (a) Maximum-likelihood phylogeny based on core-genome SNPs of the 100 isolates, 10 GenBank Haemophilus influenzae genome sequences and the reference genome, Rd_KW20. SNP groups are labelled (a–c) and CCs as defined by MLST are shaded. (b) Serotype, MLST ST and putative virulence gene profile for each isolate. Serotype and STs are designated individual colours, while for the binary virulence profiles, purple represents present, yellow absent.
Overall, SNP typing resolved the isolates into three major groups (A, B and C). Group C was a distinct group comprising entirely of Hie and Hif isolates which were more distantly related to the other encapsulated and NTHi isolates. All remaining isolates (with the exception of Hi42) clustered into either group A or B. Group A comprised entirely of NTHi (n = 33) while group B included NTHi (n = 40) and encapsulated isolates belonging to Hia, Hib, Hic and Hid.
Comparisons between NTHi in groups A and B demonstrated a notable difference; isolates possessing igaB belonged exclusively within group A and comprised 48·5% (n = 16) of these isolates. Conversely, the proportion of isolates which carried the hif gene cluster was higher in group A (18·2%, n = 6) than in group B (10%, n = 4).
Within these major groups, NTHi isolates were largely heterogeneous by SNP typing, however isolates within the same serotype and MLST ST clustered together. Isolates within these clusters were also observed to contain similar virulence gene profiles. Hi26 was the only NTHi to cluster with encapsulated isolates by SNP analysis, grouping with the Hia ST56 isolates.
SNP typing identified one isolate (Hi42) which failed to cluster with any of the three main SNP groupings. This isolate was characterised by MLST as ST1595, a newly assigned ST with two novel alleles (frdB = 177 and fucK = 120). Similarly to the other isolates, Hi42 possessed hap, pepN and iga, however was negative for all other putative virulence genes.
Antimicrobial resistance
Ampicillin-related sensitivity data were available for 92 of the 100 isolates and of these, 23 (25%) demonstrated resistance, all due to β-lactamase production. Only one of these demonstrated resistance to amoxicillin/clavulanic acid. Of these 23 isolates, 87% were NTHi (n = 20) with one isolate each of Hib, Hie and Hif. The bla TEM-1 gene was detected exclusively in each of the β-lactamase producers. No β-lactamase-negative ampicillin resistant (BLNAR) isolates were identified. No change in antimicrobial resistance was observed over time.
Additionally, 18·7% of isolates with phenotypic data (n = 14/75), all NTHi, recorded resistance to cotrimoxazole. No other significant resistances to antibiotics were observed and ResFinder did not detect the presence of any other acquired resistance genes in addition to bla TEM-1.
DISCUSSION
Rates of NTHi among invasive Hi isolates
NTHi were the most frequent cause of invasive Hi disease in QLD (72·2%), followed by Hib (11·1%) and Hif (6·9%). Similar observations of invasive Hi proportions have also been made in Spain (NTHi 86·6%) and Portugal (NTHi 77·1%) [Reference Bajanca-Lavado6, Reference Garcia-Cobos24]. Additionally, the yearly increase in the referred numbers of invasive NTHi observed for QLD has also been reported worldwide [Reference Dworkin, Park and Borchardt7–Reference Wan Sai Cheong12]. The highest NTHi rates from QLD patients were observed in the older age categories (82·3% in patients over 5 years of age), which is consistent with some observations [Reference Dworkin, Park and Borchardt7, Reference Garcia-Cobos24] though different from a study performed in Arkansas which saw a majority of NTHi invasive cases in children younger than 4 years of age [Reference O'Neill4]. This high proportion of invasive NTHi infections suggests that the lack of capsule genes does not appear to affect the ability of H. influenzae to cause invasive disease and there must be other factors which contribute to invasiveness.
Detection of putative virulence genes
All invasive QLD Hi isolates possessed pepN in addition to homologues of the iga and hap genes. These results are consistent with previous reports [Reference O'Neill4, Reference De Chiara5, Reference Kostyanev and Sechanova25–Reference Vitovski27]. Vitovski et al. [Reference Vitovski27] postulated that the presence of Iga1 protease contributes significantly to the pathogenic potential of Hi due to increased prevalence in symptomatic strains over asymptomatic strains, however as only invasive isolates were investigated in this study no further conclusions can be drawn.
The igaB gene was observed in 21·6% of invasive NTHi isolates and occurred in 12 different STs (including CCs 2 and 4), all of which shared a consistent gene profile (possession of iga, igaB, hmw genes, hap and pepN). All isolates with igaB were clustered within SNP group A. Possession of this second copy of the immunoglobulin A1 protease gene is thought to contribute to enhanced virulence [Reference Fernaays28].
We saw a high rate of hmw1A and hmw2A prevalence among NTHi isolates with 71·6% and 73% of isolates possessing these genes respectively, proportions which remained consistent within SNP groups A and B. This is higher than rates observed by Ecevit et al. and O'Neill et al. [Reference O'Neill4, Reference Ecevit15] but similar to those in Kostyanev and Sechanova [Reference Kostyanev and Sechanova25], and St. Geme et al. [Reference St. Geme29]. Consistent with other studies, no hmw genes were observed in encapsulated isolates, all of which possessed hia [Reference Ecevit15, Reference St. Geme29]. Both HMW proteins and the H. influenzae adhesin (encoded by hia) mediate the attachment of H. influenzae to human epithelial cells and they are seldom observed together in H. influenzae strains. It postulated that the possession of a particular adhesin could determine genetic origin [Reference St. Geme29], with possession of the hia gene by an NTHi strain being suggestive of evolution from an encapsulated ancestor [Reference Krasan30]. In this study, all NTHi possessing hia isolates clustered in SNP group C with the encapsulated isolates. A homologue of hia was observed in all invasive encapsulated isolates as expected, and in 17·6% of NTHi (65%, n = 13/20 of the hmw-negative isolates), lower than those observed by Ecevit et al. [Reference Ecevit15] and O'Neill et al. [Reference O'Neill4] but again similar to rates observed in Kostyanev and Sechanova [Reference Kostyanev and Sechanova25] and St. Geme et al. [Reference St. Geme29].
The hifBC target representing the pilus gene cluster was detected in 13·5% of NTHi isolates, a similar rate as observed by Ecevit et al. [Reference Ecevit15] but much less than the almost 50% rate as reported by Kostyanev and Sechanova [Reference Kostyanev and Sechanova25]. The pilus cluster was not observed in any of the QLD encapsulated isolates except for Hic, although other studies have also observed these genes in both Hib and Hif [Reference O'Neill4, Reference Ecevit15].
The sodC gene was detected in all Hie and Hif isolates and one NTHi (Hi1). This isolate was not associated with Hie or Hif by SNP or MLST; rather it was ST103, part of the largest NTHi CC1, with five other ST103 NTHi and four SLV (single-locus variants) which all additionally possessed hmw targets. Additionally, as no capsule related or IS1016 targets were detected in this isolate; it does not appear to be a capsule-deficient mutant of an encapsulated strain. It is possible that NTHi such as this isolate possess sodC through interspecies recombination with Haemophilus haemolyticus [Reference Mccrea16].
In total, six invasive NTHi isolates did not possess hmw genes, hia or hifBC, and we postulate that either Hap or other additional adhesive structures may have facilitated adherence and invasion for these isolates. These isolates did not demonstrate notable genetic relatedness by SNP or MLST.
Detection of capsule-related genes in sequenced isolates
All encapsulated isolates possessed the expected capsule gene for their known serotype, and both bexA and bexB genes in addition to the insertion sequence IS1016 (absent in one Hid isolate). No evidence of capsule-specific sequences or bexA and/or bexB were found within any of the NTHi isolates although Hi26 did possess DNA homologous to the insertion sequence IS1016 and was ST56, with a sequence type shared with Hia isolates. This isolate also possessed hia and was absent for hmw genes, similar to other IS1016-positive invasive NTHi from previous studies [Reference St. Geme29, Reference Satola, Napier and Farley31]. IS1016 has been linked with the capacity to cause severe infections [Reference Karlsson and Melhus32] and Satola et al. [Reference Satola, Napier and Farley31] also observed a statistically significant association between younger age and the presence of IS1016 in invasive NTHi. Hi26 was isolated from a 3-year-old patient. Within Hi26, the capsule region has been deleted, possibly by recombination between duplicated IS1016 copies flanking the capsule locus, but now retains a single IS1016 copy and is non-encapsulated, similar to the laboratory strain Rd_KW20. Interestingly, a previous study also demonstrated evidence of a capsule-deficient Hia variant isolated from blood at the same time of an encapsulated Hia from CSF in a meningitis patient [Reference Ohkusu, Nash and Inderlied33]. In this case it was postulated that the loss of the capsule may have been a result of antibody-driven selective pressure, which may also have been the case for this isolate.
Genetic diversity
Analysis by MLST and SNP typing demonstrated high genetic variability among the QLD NTHi isolates, a trait that has been observed by many other studies [Reference O'Neill4, Reference Bajanca-Lavado6, Reference Musser34–Reference Puig36]. No dominant NTHi clone was observed among the QLD isolates, although the most common STs were ST103 (n = 6), ST14 (n = 4) and ST12 (n = 4), with 38 unique STs, many of which have been seen in other NTHi studies [Reference Sill10, Reference Puig36].
In contrast, the encapsulated isolates largely demonstrated genetic relatedness. The STs associated with our encapsulated isolates have also been reported in the same serotypes disseminated globally [Reference Bajanca-Lavado6, Reference Sill10, Reference Puig36–Reference Ulanova and Tsang39]. No capsular switching events were evident among our isolates; the only NTHi to cluster with the encapsulated strains by MLST was Hi26 (ST56) as discussed previously.
In particular, NTHi isolates belonging within SNP group A appear to have a distinct genotype from the encapsulated isolates, suggesting a more separate population of H. influenzae than the group B NTHi isolates, which may be derived from encapsulated strains by the loss of the capsule locus through more recent evolutionary events.
Antimicrobial susceptibility testing
The only mechanism for β-lactam resistance identified in the QLD isolates was β-lactamase production; with resistance rates at 25%, within the 10–25% reported in most regions although well below rates identified in areas of Asia [Reference Van Eldere11]. Consistent with a previous global prevalence report [Reference Farrell40] the β-lactamase identified in all QLD isolates was TEM-1. Similarly to the low rates of BLNAR seen in USA and most of Europe, no BLNAR were identified from these invasive QLD isolates [Reference Van Eldere11].
Limitations of study
One limitation to this study is the lack of clinical data available to the laboratory, preventing extrapolation of disease severity to genotype. As only invasive isolates were investigated; further comments cannot be made regarding the sharing of virulence determinants between QLD invasive and non-invasive infections and future studies are warranted regarding this.
CONCLUSION
The total numbers of NTHi identified in invasive Hi disease in QLD is increasing yearly. Analysis of virulence genes could not identify any virulence determinants conclusively associated with invasive NTHi QLD isolates. Furthermore, the heterogeneity of isolates demonstrated that the QLD invasive NTHi population is genetically diverse, indicating that clonal expansion is not the reason for the increase in NTHi invasive infections in QLD. Genotyping did not identify a close relationship with encapsulated H. influenzae isolates, with only one NTHi showing evidence of a recent loss of encapsulation. In conclusion, the highly genetically diverse NTHi population identified in QLD offers a challenge to vaccine programmes and demonstrates the importance of maintaining genetic surveillance of H. influenzae isolates.
ACKNOWLEDGEMENTS
This work was supported by the Health Support Queensland Research and Development Funding. The authors would like to acknowledge Vicki Hicks and Lawrence Ariotti for their work in the identification and serotyping of referred isolates, and Helen Smith for supervision of the Clinical Reference area and for input into the study design. This publication made use of the Haemophilus influenzae MLST website (http://pubmlst.org/hinfluenzae/) sited at the University of Oxford [Reference Jolley and Maiden41]. Special thanks to Roisin Ure for collation of this database and assigning of new alleles and STs. The development of this site has been funded by the Wellcome Trust.
DECLARATION OF INTEREST
None.
