Population genetics and antibiotic susceptibility of invasive Haemophilus influenzae in Manitoba, Canada, from 2000 to 2006

Authors


  • Editor: Kai Man Kam

Correspondence: Raymond S.W. Tsang, National Microbiology Laboratory, 1015 Arlington Street, Winnipeg, MB, Canada R3E 3R2. Tel.: +1 204 789 6020; fax: +1 204 789 2018; e-mail: raymond_tsang@phac-aspc.gc.ca

Abstract

One hundred and twenty-two isolates of Haemophilus influenzae causing invasive disease were collected in Manitoba, Canada, from 2000 to 2006 and examined for serotype, biotype, sequence type (ST) by multilocus sequence typing and antibiotic susceptibility. Nonserotypeable (NST) isolates accounted for over half of the isolates collected (69 isolates, 56.6%). There were 36 serotype a, five serotype b, two serotype c, one serotype d, four serotype e and five serotype f isolates collected. The 69 NST isolates were found to be very diverse, with isolates representing six biotypes and 45 STs. The serotypeable isolates were more clonal, with each of the serotypes showing little diversity in their biotypes and STs. Of the 122 isolates, 17% were resistant to ampicillin due to β-lactamase production, 10.7% were resistant to trimethoprim-sulfamethoxazole, 1.6% were resistant to clarithromycin, 2.5% were resistant to amoxicillin-clavulanic acid and none was resistant to ciprofloxacin or moxifloxacin. Antibiotic resistance was more common in the NST strains, with 37.7% showing resistance to at least one antibiotic compared to 15% in the serotypeable strains. The results of this study suggest a shift in the epidemiology of invasive H. influenzae infections in the post-Hib vaccine era, and surveillance should include all serotypeable and NST isolates.

Introduction

Haemophilus influenzae is an obligate parasite of the human respiratory tract and causes a variety of systemic and localized infections (Jordens & Slack, 1995). One of the better known virulence factors of H. influenzae is the polysaccharide capsule, which is the antigen involved in serotyping of strains (Pittman, 1931) as well as the target for vaccine immunity (Makela et al., 1977; Peltola et al., 1977). Margaret Pittman in 1931 developed specific antisera that recognized six serotypes (a–f) of H. influenzae based on antigenic specificities of the different capsule structures (Pittman, 1931). Genetic analysis of the capsular operon in the different serotypes confirmed the presence of serotype-specific capsule polysaccharide synthesis (cps) genes (Kroll et al., 1989). Based on the biochemical reactions to detect production of indole, urease and ornithine decarboxylase, H. influenzae is subdivided into eight biotypes (Kilian, 1976). Correlations between biotypes and serotypes as well as sources of H. influenzae have been described (Kilian, 1976). Besides biochemical differences, strains of H. influenzae can also be grouped by the method of multilocus sequence typing (MLST) (Meats et al., 2003), which is based on assigning allele numbers to partial sequences of seven house-keeping enzyme genes. Strains with at least four identical alleles are classified as a clonal complex.

Prior to the introduction of vaccines against the serotype b H. influenzae (Hib), most invasive H. influenzae infections were due to Hib (Wenger et al., 1992; Falla et al., 1993). Since the introduction of the Hib conjugate vaccine, infections due to Hib have plummeted to an all time low with the hope that Hib disease can be eliminated (Bath et al., 2002). However, with the decrease in the number of invasive Hib infections, invasive infections due to other serotypes [including the nonserotypeable (NST)] have become more apparent (Cerquetti et al., 2000; Bajanca et al., 2004; Campos et al., 2004). To understand the evolving nature of invasive H. influenzae infections, we carried out a retrospective study to examine isolates from invasive disease cases. This represents an in-depth study of the genetics (by MLST) and antibiotic susceptibility of H. influenzae strains collected in Manitoba, Canada, in the post-Hib vaccine era.

Materials and methods

Bacterial strains and biochemical testing

Isolates of H. influenzae from normally sterile body sites (cerebrospinal fluids, blood, joint fluids, etc.) were collected from the clinical microbiology laboratories of the Health Sciences Centre and the St Boniface Hospital in Winnipeg, Manitoba, as well as from the Cadham Provincial Public Health Laboratory. Fifty-two of the 122 isolates collected in this study have been described before (Tsang et al., 2006). Control strains of Hib, Rm114 and Eagan, were obtained from Dr Derrick Crook and Dr Richard Moxon, John Radcliffe Hospital in Oxford, UK.

The identities of the isolates were confirmed by standard biochemical tests (Kilian, 2003) and biotypes were assigned according to current nomenclature (Kilian, 1976).

Serotype determination

Serotyping was done by slide agglutination using antisera from two commercial sources (Difco, Oakville, Ontario, Canada; Denka Seiken, Tokyo, Japan). Detection of serotype-specific capsular polysaccharide synthesis genes by PCR was done using primers and procedures described by Falla et al. (1994).

Multilocus sequencing typing

Multilocus sequence typing (MLST) was carried out according to a previously described method (Meats et al., 2003), and the assignment of sequence types (STs) was done using the H. influenzae MLST website (http://haemophilus.mlst.net/).

Antibiotic susceptibility testing

The production of β-lactamase was detected using DrySlide nitrocefin (BBL, Becton Dickinson, Oakville, Ontario, Canada). Susceptibility towards antibiotics was determined by the disk diffusion method described by the Clinical and Laboratory Standards Institute (CLSI), formerly known as the National Committee for Clinical Laboratory Standards (NCCLS) (National Committee for Clinical Laboratory Standards, 2003). Overnight cultures of H. influenzae were resuspended in sterile phosphate-buffered saline (PBS) and adjusted to a 0.5 McFarland turbidity standard before being plated onto Haemophilus test medium agar plates (Oxoid, Nepean, Ontario, Canada). After applying antibiotics disks, plates were incubated at 35C with 5% CO2 for 18–24 h. The following antibiotics (Oxoid, Nepean, Ontario, Canada) were tested: ampicillin (2 and 10 μg), amoxicillin-clavulanic acid (30 μg), cefaclor (30 μg), ceftriaxone (30 μg), chloramphenicol (30 μg), trimethoprim-sulfamethoxazole (25 μg), ciprofloxacin (5 μg), moxifloxacin (5 μg), and clarithromycin (15 μg). Detection of β-lactamase negative ampicillin-resistant strains was accomplished by testing strains against two concentrations of ampicillin (Karpanoja et al., 2004). Zones of inhibition were measured and interpreted according to CLSI guidelines (National Committee for Clinical Laboratory Standards, 2003). Haemophilus influenzae ATCC 49247 was used as a control in each experiment.

Results

The serotype distribution of 122 H. influenzae isolates collected from individual invasive disease cases in Manitoba from 2000 to 2006 are described in Table 1. Fifty-two of these isolates had been described before (Tsang et al., 2006). Other than an apparent increase in the number of NST strains in the last 3 years, no other trends were observed with any of the serotypeable strains during this study period. Age and sex of these 122 cases and the incidence rates of invasive H. influenzae disease caused by Hib and non-Hib were described separately (Tsang et al., 2007). Besides describing the age and sexes of the cases involved, we also documented that the incidence rates of invasive H. influenzae disease due to non-Hib strains were similar to Hib rates found in the pre-Hib vaccine era (Tsang et al., 2007).

Table 1.   Distribution of invasive Haemophilus influenzae serotypes by year in Manitoba, Canada
Haemophilus influenzae serotypesYear of isolation
2000200120022003200420052006All years
Serotype a386647236
Serotype b02001025
Serotype c00011002
Serotype d10000001
Serotype e00001214
Serotype f01110115
Nonserotypeable884313191469
All serotypes12191111202920122

Haemophilus influenzae serotype a (Hia)

All 36 invasive Hia isolates were found to belong to biotype II, and 35 of them were of the sequence type ST-23. The remaining isolate was designated ST-397 and shared six of its seven MLST house-keeping gene alleles with ST-23, and was therefore a member of the ST-23 clonal complex (Table 2). Only one of these 36 isolates was found to produce β-lactamase and was resistant to ampicillin (Table 3). The other 35 isolates did not produce β-lactamase and were fully sensitive to the panel of antibiotics tested.

Table 2.   MSLT allelic profiles of serotypeable Haemophilus influenzae strains from invasive disease cases in Manitoba from 2000 to 2006
Haemophilus influenzae serotypeNo. of isolatesMLST allelic profileSequence type
adkatpGfrdBfucKmdhpgirecA
  1. Allelic differences between strains within each serotype are highlighted in bold.

Serotype a35131652311723
113165231185397
Serotype b331144547895
110144343844
11014458578190
Serotype c27116861699
Serotype d1515109851110
Serotype e218632210281269
1186314102812386
1186352102812384
Serotype f522191111221915124
Table 3.   Antibiotic susceptibility pattern of invasive Haemophilus influenzae isolates by serotype
Haemophilus
influenzae
serotype
No. of
isolates
β-Lactamase
production
Antibiotic
susceptibility pattern*
  • *

    Antibiotics tested: ampicillin (Amp) 2 μg and 10 μg; ceftriaxone (Cro) 30 μg; chloramphenicol (C) 30 μg; trimethoprim-sulfamethoxazole (Sxt) 25 μg; amoxicillin-clavulanic acid (Amc) 30 μg; cefaclor (Cec) 30 μg; ciprofloxacin (Cip) 5 μg; moxifloxacin (Mxf) 5 μg; clarithromycin (Clr) 15 μg.

Serotype a35Sensitive to all
1+Amp resistant
Serotype b2Sensitive to all
2+Amp resistant
1+Amp, C, Sxt resistant
Serotype c2Sensitive to all
Serotype d1Sensitive to all
Serotype e1Sensitive to all
1+Amp resistant
1+Amp, Clr resistant
1+Amp, Amc resistant
Serotype f4Sensitive to all
1Clr resistant
Nonserotypeable42Sensitive to all
10+Amp resistant
2+Amp, Amc resistant
2+Amp, Sxt resistant
8Sxt resistant
2Amp intermediate, Sxt resistant
2Amp intermediate, Cec resistant
1Amp, Cec intermediate

Haemophilus influenzae serotype b (Hib)

The five Hib isolates all belonged to biotype I and formed a cluster of related sequence types belonging to one clonal complex (Table 2). Three Hib isolates were found to produce β-lactamase and were resistant to ampicillin; and in one of these three Hib isolates, resistance to trimethoprim-sulfamethoxazole and chloramphenicol was also found.

Haemophilus influenzae serotypes c, d, e and f (Hic, Hid, Hie and Hif)

All four Hie isolates as well as the single Hid isolate belonged to biotype IV. Both Hic isolates belonged to biotype II. Among the five Hif isolates, four were biotype I and one was biotype II. Clonal analysis by MLST revealed that strains from each serotype (c, d, e and f) represented a distinct clonal population, with unique STs that shared no common house-keeping gene alleles (Table 2). Resistance to antibiotics appeared to be more common in Hie than in Hif isolates; and no resistance to antibiotics was found in the small number of Hic or Hid isolates tested (Table 3).

Nonserotypeable Haemophilus influenzae (NST)

There were more nonserotypeable strains (69 isolates) than all the serotypeable strains combined (53 isolates). The NST H. influenzae isolates from these 69 invasive disease cases formed a very heterogeneous population with six different biotypes (I–VI) being represented. The most common biotypes found were biotype II (39 isolates, or 56.5%) and biotype III (18 isolates, or 26.1%). The diversity of the NST H. influenzae isolates was confirmed by MLST as 45 different STs were found among this group of organisms (Fig. 1). The most commonly found ST was ST-14 (represented by eight isolates), followed by ST-3, ST-145 and ST-156 (each represented by four isolates). Thirty-three isolates had unique STs, and many were not related to each other based on their house-keeping gene allelic profile.

Figure 1.

 Dendrogram showing the multilocus sequence typing results and antibiotic susceptibility patterns for 69 nonserotypeable Haemophilus influenzae isolates from 2000–2006 in Manitoba, Canada.

Fourteen isolates (20%) were found to produce β-lactamase, and were resistant to ampicillin. Among the β-lactamase positive strains, two showed additional resistance to amoxicillin-clavulanic acid, two also had resistance to trimethoprim-sulfamethoxazole. Overall, 26, or 37.7%, of the 69 isolates showed resistance to at least one antibiotic.

Discussion

Our data confirmed the importance of non-Hib H. influenzae strains in causing invasive disease in Manitoba in the post-Hib vaccine era. We confirmed a proportional shift in non-Hib strains causing invasive H. influenzae disease and that the prevalence of invasive H. influenzae disease may be increasing, approaching the Hib rates found in the prevaccine era (Tsang et al., 2007). As the total number of Hic, Hid, Hie, and Hif strains accounted for only 10% of all invasive H. influenzae isolates, capsule replacement by these four serotypes in the post-Hib vaccination era does not appear to contribute significantly to the changes in the epidemiology of invasive H. influenzae disease.

Animal studies with isogenic mutants transformed with serotype-specific capsule-associated DNA from all six serotypes (a–f) had shown that although serotype b was the most virulent serotype, serotype a was found to be more virulent than serotypes c, d, e, and f (Zwahlen et al., 1989). In line with this observation is that the serotype a capsule structure is more similar to that of serotype b than to other H. influenzae serotypes. Both the Hia and Hib capsules contain the five-carbon sugar ribitol. In Hia, ribitol is linked to glucose (Branefors-Helander et al., 1977) whereas in Hib, ribitol is linked to ribose (Crisel et al., 1975).

Although we may be able to explain why Hia is a significant cause of invasive H. influenzae disease in the post-Hib era, there are no obvious reasons for NST strains to cause a large number of invasive disease cases. The highly heterogeneous nature of the NST strains (in terms of their biotypes, DNA fingerprints, and MLST types) isolated from the invasive disease cases do not suggest that one or a few virulent strains have emerged to cause invasive disease. Genetic characterization of NST H. influenzae strains has suggested a panmictic structure (Musser et al., 1986). More recent data also suggest that this group of H. influenzae may contain a supra-genome that will offer strains in different environmental niches the set of genes required for survival or cause infections (Shen et al., 2005). No single age group appears to be more vulnerable to infections by the NST H. influenzae strains. However, lack of clinical data of these cases prevented us from drawing any further conclusion, and further prospective studies may be required that should focus on pre-existing medical conditions [such as presence of chronic diseases such as diabetes, asthma, chronic obstructive pulmonary disease (COPD), and cancer] of the patients suffering from invasive disease due to NST H. influenzae strains. NST H. influenzae strains are frequent colonizers of the upper respiratory tract and are known to cause exacerbations of COPD (Chin et al., 2005). Therefore, NST H. influenzae pneumonia with secondary bacteraemia may account for some of our cases.

Using MLST, a unique relationship between serotypes and their population structure was found. Each serotype appeared clonal and belonged to a unique clonal complex. STs found within a serotype were related in at least five of the seven house-keeping gene alleles sequenced. On the other hand, strains that belong to different serotypes were found to share no single common house-keeping gene allele (Table 2). This association between serotypes and clonal complexes has been previously demonstrated (reviewed by Pennington, 1993). As such, our data (Table 2) confirmed no capsule switching as we have not found any non-Hib strains in the post vaccination era with STs typically found in Hib strains in the prevaccine era. Also, most of the STs found in the serotypeable strains from Manitoba appeared to have been described in the MLST website, which also included all the STs described in the paper by Meats et al. (2003). There were only three new STs found in this study and they involved one Hia (ST-397) and two Hie (ST-384 and ST-386) strains. On the other hand, NST H. influenzae strains were found to be highly diverse and their population structure has been described as panmictic (Musser et al., 1986).

Most antibiotic susceptibility data on H. influenzae focused primarily on respiratory isolates. This study focuses on invasive isolates of both serotypeable and NST strains. There was no difference between the serotypeable and NST H. influenzae strains with regard to their ability to produce β-lactamase or resistance to ampicillin and amoxicillin-clavulanic acid. However, 12, or 17.4%, of NST vs. only 1, or 1.9%, of 53 serotypeable H. influenzae were found to have resistance to trimethoprim-sulfamethoxazole. No resistance was identified to the fluoroquinolones, ciprofloxacin and moxifloxacin, commonly used agents for many infections, particularly pneumonia. The overall rate of β-lactamase producing H. influenzae was 17%. This appears to reflect the general decrease in the rate of production of β-lactamase in H. influenzae found in two other previous studies; 24.2% of strains were β-lactamase positive in a 1997–1998 study and 18.2% were positive in a 2001–2002 study (Zhanel et al., 2003; Heilmann et al., 2005). Although β-lactamase negative ampicillin-resistant (BLNAR) strains were not found in this study, three β-lactamase producing amoxicillin-clavulanic acid-resistant isolates were detected, with an overall rate of 2.5%. Whether these isolates have decreased affinity toward ampicillin due to altered penicillin-binding proteins or have extended spectrum β-lactamase activities requires further investigation. In addition, five β-lactamase negative NST isolates demonstrated a decreased susceptibility toward ampicillin (intermediate by disk diffusion test). Whether these belong to the class of low BLNAR strains (Qin et al., 2007) would require further determination of their exact minimum inhibitory concentrations (MIC)s and their binding affinity to ampicillin. The overall rate of resistance to clarithromycin (two of 122 isolates, or 1.6%) may also suggest a slight increase from a previous survey of respiratory H. influenzae in Canada (Zhanel et al., 2003).

In summary, this study highlighted the importance of Hia and NST strains in causing invasive H. influenzae disease in Manitoba, Canada. Although less frequent, serotypes c, d, e, and f have also been isolated from invasive disease cases. Therefore, surveillance systems for invasive H. influenzae disease should be expanded to include all H. influenzae regardless of their capsular antigen status (serotypeable and NST) as long as they are recovered from normally sterile body sites. Inclusion of clinical data in future surveillance system will also help to provide a better understanding of the evolving nature of the H. influenzae population as well as the hosts that are susceptible to invasive disease by these microorganisms.

Our data also defined the clonal nature of the serotypeable and NST invasive H. influenzae strains in Manitoba, Canada. Documentation of their genetic characteristics, as represented by their sequence types, will facilitate any future investigations of potential capsule switching among H. influenzae patient and/or carrier isolates. The data on antibiotic susceptibility also highlighted the need for continue surveillance and further investigation into the prevalence of clarithromycin and BLNAR as well as the BLPAR strains in Manitoba and other parts of Canada.

Acknowledgements

R.S.W. Tsang was supported by Health Canada's Genomics R&D Funding for the genetic studies of Haemophilus influenzae.

Ancillary