Since the introduction of Haemophilus influenzae type b (Hib) conjugate vaccines, meningitis caused by serotypes other than Hib has gained in importance. We conducted active hospital-based surveillance for meningitis over an 11-year period in Salvador, Brazil. H. influenzae isolates were serotyped and analyzed by polymerase chain reaction, pulsed-field gel electrophoresis, and DNA sequencing to identify strains with a specific deletion (IS1016) in the bexA gene (IS1016-bexA). We identified 43 meningitis cases caused by non-type b H. influenzae: 28 (65%) were caused by type a (Hia), 9 (21%) were caused by noncapsulated strains, and 3 (7%) each were caused by types e and f. Hia isolates clustered in 2 clonal groups; clonal group A strains (n = 9) had the IS1016-bexA deletion. Among children <5 years of age, meningitis caused by Hia from clonal group A had higher case-fatality than meningitis caused by clonal group B. Despite small numbers, these results indicate that the presence of the IS1016-bexA deletion is associated with enhanced virulence in non-type b H. influenzae.
Introduction of Haemophilus influenzae type b (Hib) conjugate vaccines into childhood immunization programs has dramatically reduced the incidence of Hib meningitis in countries using Hib vaccines [1–3]. Hib conjugate vaccines are highly efficacious against invasive Hib disease , decrease Hib carriage among vaccinated children, and reduce transmission and invasive disease among nonimmunized children .
Hib conjugate vaccines do not prevent H. influenzae disease caused by other serotypes, raising the potential for the emergence of H. influenzae disease due to virulent organisms with non–type b capsules [6–11]. Detection of meningitis due to non-type b H. influenzae has increased following widespread use of Hib conjugate vaccines as a result of improved surveillance and use of molecular techniques, which have reduced the serotyping errors associated with slide agglutination [12–14]. Molecular methods have also been used to identify genetic elements in invasive non-type b H. influenzae isolates, including presence of a partial deletion of IS1016 in the bexA gene commonly found in Hib isolates. The IS1016-bexA deletion is a putative virulence factor that has been identified in invasive Hia isolates from patients with severe disease in The Gambia and the United States [9, 10, 15], but not in all areas where Hia strains have been isolated [11, 16]. Acquisition of virulence factors from Hib strains could possibly lead to the emergence of non-type b H. influenzae disease.
Previously, we reported a transient increase in meningitis due to H. influenzae serotype a after introduction of Hib conjugate vaccine in Salvador, the third largest urban center in Brazil [17, 18]. Clinical outcomes of meningitis cases due to non-type b H. influenzae were similar to those of cases due to Hib [17, 18]. To investigate the role of the IS1016-bexA deletion in clinical outcomes of meningitis cases due to non-type b H. influenzae, we analyzed data from 11 years of active, hospital-based meningitis surveillance in Salvador, Brazil.
Surveillance. Meningitis is a nationally notifiable disease in Brazil, with mandatory reporting of all suspect meningitis cases to public health authorities. We conducted active surveillance for meningitis among patients admitted to Couto Maia Hospital in Salvador, Brazil. According to state health guidelines, all suspected cases of meningitis in the region are referred to Couto Maia Hospital for diagnostic procedures, including lumbar puncture and examination of cerebrospinal fluid. Couto Maia Hospital accounted for 98% of reported meningitis cases among persons from the metropolitan area of Salvador during the study period .
We analyzed data for H. influenzae meningitis cases identified from 9 March 1996 through 8 September 2007. A case of H. influenzae meningitis was defined as a patient who had: (1) clinical presentation of meningitis, characterized by fever, meningismus, and altered mental status; (2) abnormal cerebrospinal fluid examination; and (3) cerebrospinal fluid or blood culture positive for H. influenzae. The study team reviewed laboratory records 5 days a week to identify new culture isolations of H. influenzae. Patients were enrolled in the study according to informed consent procedures approved by the Institutional Review Boards of the Oswaldo Cruz Foundation, Brazilian Ministry of Health, and the New York–Presbyterian Hospital (New York, New York). We used a standardized data entry form to collect information on demographic characteristics, clinical presentation, laboratory results, and outcome from the patient's medical records. Number of doses of Hib conjugate vaccine received prior to hospitalization and dates of vaccination were obtained from patient immunization records.
Strain identification and serotyping. H. influenzae was identified by Gram stain morphology and growth requirement for hemin and nicotinamide adenine dinucleotide. Commercial antiserum (Becton Dickinson) was used to determine capsular serotype. Each isolate was tested for slide agglutination with the complete panel of type a-specific to type f-specific antisera (Becton Dickinson) and a saline control. A semi-nested polymerase chain reaction (PCR) method was used to amplify serotype- specific and nonspecific DNA sequences from the H. influenzae capsular loci . Isolates were defined as noncapsulated if agglutination was not observed with the 6 type-specific antisera and if PCR capsular loci sequences conserved among serotypes were not detectable by PCR .
Pulsed-field gel electrophoresis characterization (PFGE). H. influenzae non-type b clinical isolates were examined by PFGE after digestion of bacterial DNA with Sma I (New England Biolabs), as previously described [21, 22]. The Sma I fingerprints were analyzed using GelCompar II software (Applied Maths). A 1.5%-band position tolerance was used for gel comparisons. Cluster analysis was performed using the un-weighted-pair-group method, and the relatedness between isolates was interpreted according to the criteria of Tenover .
Identification of the IS1016-bexA partial deletion and sequencing.H. influenzae non-type b isolates and a random sample of 20 Hib isolates were evaluated by PCR for identification of a partial deletion of the bexA gene, using the IS1016 and bexA primers as previously described . For DNA sequencing, PCR products were purified with the QIAquick PCR purification kit (Qiagen) and subjected to sequence analysis. The DNA sequences from both strands were edited, assembled, and aligned using MEGA4 and BioEdit software. The sequences were compared with those of the Hib strains AF549213 , S62752 , and the type a strain DQ086152 , available in the NCBI Gene bank.
Multilocus sequence typing (MLST). MLST was performed for 2 H. influenzae type a isolates that were randomly selected among the isolates which had and did not have the IS1016- bexA deletion. Chromosomal DNA was extracted using a Qiagen genomic Kit (Qiagen). PCR was used to amplify 450-base pair (bp) internal fragments of 7 housekeeping genes (adk, atpG, frdB, fucK, mdh, pgi, and recA), according to previously described methods . Sequences were submitted to the online MLST database (http://www.mlst.net), which in turn assigned alleles at each locus and a sequence type.
Statistical analysis. Data were entered and analyzed using EpiInfo software, version 3.3.2 (Centers for Disease Control and Prevention). Fisher's exact test and the Wilcoxon rank-sum test were used for comparison of proportions and continuous data, respectively. A significant difference was defined by a 2-tailed P-value < .05.
Mean annual incidence of H. influenzae meningitis was compared for the period prior to introduction of Hib vaccination (March 1996 through July 1999) and after Hib vaccine introduction (August 1999 through September 2007). Incidence was calculated for the metropolitan area of Salvador by dividing the number of cases among residents of metropolitan Salvador by the estimated population from the 2000 national census .
During the study period, we identified 615 cases of H. influenzae meningitis. Among the 573 cases (93%) for which an isolate was serotyped, 43 episodes (8%) were caused by H. influenzae non-type b strains (Table 1). The majority of H. influenzae non-type b isolates were type a (28 isolates; 65%), followed by noncapsulated (9 isolates; 21%), type e (3 isolates; 7%), and type f (3 isolates; 7%). The proportion of H. influenzae meningitis cases caused by a non-type b isolate increased from 2% (8 of 424) to 23% (35 of 149) after the introduction of routine Hib immunization (P < .001). This increase was largely explained by the 91% reduction in the incidence of Hib meningitis between the pre- and post-vaccine periods (from 2.45 to 0.24 cases per 100,000 population; P < .001). The incidence of meningitis due to non-type b H. influenzae increased after the introduction of the Hib conjugate vaccine, mainly because of an increase in disease due to Hia. Meningitis cases due to Hia did not cluster spatially with respect to the neighborhood of residence during pre- and postvaccine periods.
Hia and Hib meningitis occurred mainly among children <5 years of age, whereas meningitis due to H. influenzae types e, f, and noncapsulated strains occurred in older age groups (Table 2). Case-fatality was also higher for Hia and Hib meningitis cases than it was for meningitis cases due to other serotypes (Table 2). The age group distribution and case fatality rate for H. influenzae type a cases did not differ between the pre- and postvaccine period.
We were able to obtain information on immunization status for 26 (74%) of the 35 meningitis cases due to non-type b H. influenzae identified in the postvaccine period. Although 75% (13 of 17) of the patients with cases due to H. influenzae type a isolates had received 2 or 3 Hib vaccine doses, only 11% (1 of 9) of the cases due to H. influenzae type e, f, and noncapsulated isolates received the same number of Hib vaccine doses (P < .01).
PFGE analysis for the 43 H. influenzae non-type b isolates discriminated 15 distinct patterns (Figure 1). The 28 H. influenzae type a isolates had 2 different patterns, cluster A (9 isolates), and cluster B (19 isolates), whereas H. influenzae types e, f, and noncapsulated strains were heterogeneous (Figure 1). MLST analysis determined that PFGE clusters A and B corresponded to sequence type (ST) 4 and 23, respectively. PCR analysis identified the 339-bp IS1016-bexA partial deletion product in 9 of the 43 H. influenzae non-type b isolates. All of the 9 H. influenzae isolates containing the IS1016-bexA deletion were serotype a and belonged to PFGE cluster A (ST4). Among the 28 Hia isolates, 5 and 23 were isolated during the pre- and postvaccine periods, respectively. The proportion of Hia isolates with the IS1016-bexA deletion was 40% (2 of 5) and 30% (7 of 23) in the pre- and postvaccine periods, respectively, and this difference was not statistically significant (P = .65).
Patients with meningitis cases caused by Hia strains belonging to cluster A or B were similar with respect to sex, age, and characteristics of cerebrospinal fluid (Table 3). However, the case-fatality rate for patients with meningitis caused by Hia strains that had the IS1016-bexA deletion was 33% (3 of 9 patients died), compared with 5% (1 of 19) for patients with cases caused by Hia strains with complete IS1016-bexA (P = .06) (Table 3). Among children <5 years of age with H. influenzae type a meningitis, 38% (3 of 8) of individuals with isolates that contained the IS1016-bexA deletion died, whereas none of the 16 patients with isolates that did not contain the IS1016-bexA deletion died (P = .03) (Table 3).
Sequencing of the PCR products confirmed the presence of an IS1016-bexA deletion in the 9 H. influenzae type a ST4 isolates. The size and location of the deletion, as well as the flanking region sequences, was identical to that previously reported for an invasive serotype a strain that was isolated in Georgia in 2005, except for 1 nucleotide in position 98 (GenBank accession number DQ086152) (Figure 2). However, the sequence of the regions flanking the IS1016-bexA deletion for the ST4 isolates differed at 4 nucleotide sites from corresponding sequences for 2 previously reported Hib strains (GenBank accession numbers AF549213 [HI 1007-Georgia] and S62752 [RM 7004-Gambia]) and 3 of 4 Hib stains isolated during surveillance in Salvador. One Hib strain isolated from an individual in Salvador had a flanking region sequence that differed at only 1 nucleotide from the corresponding sequence in serotype a ST4 isolates (Figure 2).
Widespread use of Hib conjugate vaccines has substantially reduced the incidence of Hib meningitis [1–3, 28, 29], resulting in increased awareness of meningitis due to other H. influenzae serotypes [7, 8, 11]. Because Hib conjugate vaccines are effective in reducing Hib nasopharyngeal carriage [30, 31], it was hypothesized that non-type b strains could potentially occupy the niche left by Hib and consequently increase the risk of invasive disease caused by non-type b strains. To date, however, there has been little evidence of a substantial replacement of Hib disease by disease caused by other serotypes, a phenomenon known as serotype replacement [17, 32].
Among the capsulated H. influenzae strains that are not type b, type a has the capsular polysaccharides most closely related to those of type b. In animal challenge studies, reports have found that Hia is the most virulent capsulated H. influenzae after Hib . The H. influenzae type a meningitis cases from this study occurred among similar age groups and had case-fatality rates similar to those for Hib meningitis. In contrast, cases due to serotype e, serotype f, and noncapsulated serotypes occurred at older ages and tended to have a better prognosis. These findings are consistent with prior clinical and epidemiological characterizations of invasive disease due to H. influenzae non-type b and support the hypothesis that type a isolates are the most virulent capsulated H. influenzae serotype after type b . Although H. influenzae type a invasive infections typically occur in healthy children [9, 10, 12, 16, 34], infections due to serotype e, serotype f, and noncapsulated serotypes mostly occur among adults with underlying conditions, such as cancer [8, 35, 36].
In this study, we found that patients with H. influenzae type a meningitis had an increased risk of death when the IS1016-bexA partial deletion was present in the clinical isolate. The association did not appear to be confounded by other prognostic factors, such as patients’ age and disease duration prior to hospitalization. This finding is both plausible and analogous with what is known about virulence factors for Hib, for which the IS1016-bexA deletion stabilizes duplicated loci and leads to increased production of capsular polysaccharide [25, 33]. Hib capsular loci amplification has been found to inhibit complement-mediated bacteriolysis and opsonization . Capsular amplification and the IS1016-bexA deletion have been identified in Hia invasive isolates [9, 10, 12, 15]. However, this study provides the first evidence, to our knowledge, for the significant association between the IS1016-bexA deletion and poor clinical outcome from Hia invasive disease.
However, the IS1016-bexA partial deletion was present in a minority of the H. influenzae type a isolates (9 of 28 isolates). Other investigations have also identified isolates of H. influenzae type a causing invasive infections resembling Hib invasive disease in the absence of the IS1016-bexA partial deletion [11, 16]. Additional studies in other geographical settings and with larger sample sizes are warranted to confirm the role of the IS1016-bexA deletion as a virulence factor in H. influenzae type a invasive disease. Furthermore, we did not evaluate whether the presence of the IS1016-bexA deletion was associated with neurological sequelae, hearing impairment, or other markers of disease severity. Finally, further studies are needed to determine whether clinical isolates with the IS1016-bexA deletion exhibit enhanced virulence in animal models for Hia infection.
Results of this study suggest that Hia strains causing meningitis in Salvador have been stable over time. Sequence type 23 has been isolated in Malaysia, Canada, and New Guinea [11, 26], which suggests worldwide spread of these clones. Interestingly, sequence type 4, previously isolated in Kenya and The Gambia , was the first non-type b strain identified as having the IS1016-bexA partial deletion . In addition, Sill et al  described a Canadian case of H. influenzae type a invasive disease due to an ST4 strain containing the IS1016-bexA partial deletion . This isolate was closely related on the basis of PFGE analysis to 2 Hia strains possessing the IS10116-bexA deletion that were isolated from patients with invasive disease in Georgia . Future studies are needed to investigate whether the ST4 clone is entirely responsible for the global spread of H. influenzae type a strains containing the IS1016-bexA partial deletion. These findings highlight the need to continue surveillance for H. influenzae invasive disease to monitor for the potential emergence of non-type b H. influenzae virulent clones.
We thank the study patients and their families; the clinical, laboratory and administrative staff of Hospital Couto Maia, especially Ana MariaMaia and Neide Oliveira Silvap; Ricardo Martinez and Tatiana Lobo, for their participation in data collection and processing; Neci Ivo Ramos, Nilda Lúcia Nunes Ivo, Maria Auxiliadora Macedo de Lima Machado, and Helena Macedo, for providing information on the Hib immunization programand meningitis case notifications; Hermes P. da Silva Filho, for support on the sequence analysis; and Brendan Flannery, for manuscript review.