Molecular Epidemiological Analysis of the Changing Nature of a Meningococcal Outbreak following a Vaccination Campaign

Abstract

A serogroup C meningococcal outbreak that occurred in an Israeli Arab village led to a massive vaccination campaign. During the subsequent 18 months, new cases of type B Neisseria meningitidis infection were revealed. To investigate the influence of vaccination on bacteriological epidemiology, bacteria were isolated from individuals at the outbreak location, patients with several additional other sporadic cases, and patients involoved in another outbreak. Haploid bacterial genomic DNA was mixed with a consensus PCR product to form a heteroduplex state that enabled multilocus sequence typing (MLST) to be combined with denaturing high-performance liquid chromatography (DHPLC) for a novel high-throughput molecular typing method called MLST-DHPLC. A 100% correlation was found to exist between the sequencing by MLST alone and the MLST-DHPLC method. Independent molecular typing by repetitive extragenic palindromic PCR discriminated the neisserial clones as well as the MLST-DHPLC method did. The occurrence of type B N. meningitidis in the postvaccination period might be attributed to the selection pressure applied to the bacteria by vaccination, suggesting a possible unwarranted outcome of vaccination with the quadrivalent vaccine for control of a serogroup C meningococcal outbreak. This is the first time that DHPLC has been applied to the genotyping of bacteria, and it proved to be more efficient than MLST alone.

Meningococcal outbreaks are one of the most feared public health emergencies (8). Recent advances in molecular typing methods allow a better understanding of the nature of these outbreaks. In the Israeli Arab village of Tamra, 16 clinical meningococcal infections occurred in a population of 25,000 people during the period from June 1996 to September 1999. All the patients were hospitalized in the Rambam Medical Center in Haifa, Israel. Cultures for 5 of the 10 cases that occurred during the first 21 months were found to be positive for Neisseria meningitidis; all isolates were classified as serogroup C. Three of these cases occurred over a 2-month period, for a primary attack rate of 12 per 100,000 population. In February 1998, the Ministry of Health intervened by vaccinating the village's entire population with a quadrivalent meningococcal vaccine. In the postvaccination stage, six additional cases were reported over a period of 19 months. Cultures for four cases were found to be positive for N. meningitidis; all were classified as serogroup B. One patient, in whom two cases were reported during the outbreak period, was shown to completely lack the C7 component of the complement system (1a). Antimicrobial prophylaxis with rifampin was administered to all family members.

The epidemiological features of meningococcal outbreaks and their relationship to different vaccinations are an issue of tremendous importance and are being investigated for safety and efficacy. Most cases of meningococcal disease are caused by a limited number of groups of genetically related bacteria within serogroups B and C, which have been referred to as hyperinvasive lineages or hypervirulent strains (2). At certain intervals, hyperinvasive lineages arise and spread locally and globally. For serogroup B and C meningococci, multilocus enzyme electrophoresis studies have shown that several complexes of genetically related but not identical isolates can be recovered from several patients in a specific outbreak (3). The members of such complexes often change antigenically (27), while they retain distinctive epidemiologies and pathologies, thus creating problems for the development of vaccines and the identification of hyperinvasive strains by serology. A good example of such a spread is that of the electrophoretic type (ET) 5 (ET-5) complex, which mainly contains serogroup B meningococci, and the ET-37 complex, which mainly contains serogroup C meningococci. Both complexes were identified as spreading globally, causing elevated rates of meningococcal infections in a number of countries (2).

The Advisory Committee on Immunization Practices recommends the use of mass vaccination to control outbreaks of serogroup C meningococcal disease (4). The United Kingdom was the first country to introduce the meningococcal C conjugate vaccines into the routine infant vaccination schedule (16). However, there is a growing concern about the use of meningococcal polysaccharide vaccines due to their ineffectiveness against serogroup B and the possibility of strain selection or capsular switching (16, 21).

We conducted the study described here in order to explore the molecular epidemiology of the Tamra outbreak and the influence of the vaccination on the outbreak. This investigation also provided us with an opportunity to enhance the development of a high-throughput molecular typing strategy that combines the highly informative multilocus sequence typing (MLST) method (12) with the speed of denaturing high-performance liquid chromatography (DHPLC). We used MLST and DHPLC in combination (MLST-DHPLC) for the molecular typing of the different bacteria and compared the results obtained by that method to those obtained by MLST alone and those obtained by the repetitive extragenic palindromic PCR (rep-PCR) method (9, 24-26, 28, 29). Scanning for mutations or polymorphisms by DHPLC involves subjecting PCR products to ion-pair reverse-phase particles. Under conditions of partial heat denaturation within a linear acetonitrile gradient, heteroduplexes that form in PCR samples with internal sequence variations display reduced column retention times relative to those of their homoduplex counterparts. The elution profiles for such samples are distinct from those for samples with homozygous sequences, making the identification of samples harboring polymorphisms or mutations a straightforward procedure (14, 22, 23). We believe that this is the first time in which a method that uses a combination of MLST and DHPLC has been applied to the analysis of bacterial genomes.

MATERIALS AND METHODS

Outbreak definition.

An outbreak was defined as the occurrence of three or more confirmed cases of serogroup C meningococcal disease over a period of 3 months or less, with a resulting primary attack rate of at least 10 cases per 100,000 people (4). A patient with clinical meningococcal disease was defined as a patient who had meningitis and a petechial rash or a clinical picture consistent with meningococcemia. A patient with a laboratory-confirmed case has a blood or cerebrospinal fluid sample that is culture positive for N. meningitidis.

N. meningitidis isolates were obtained from several sources. All nine isolates from the village of Tamra were obtained for molecular typing. Thirty more isolates from patients with sporadic cases of meningococcal disease and isolates from two different outbreaks in Israel were also typed in order to ensure the validity of the new typing method (MLST-DHPLC). Patients with sporadic cases were matched by age and time of disease to patients from whom the Tamra outbreak isolates were recovered but were from different geographic areas. Two of the sporadic cases were caused by serogroup W isolates. Two outbreaks caused by type B isolates from the north of Israel (Dir el Asad [20]) and the south of Israel (Netzarim [unpublished data]) were also analyzed. The last group of strains was from another Arab village, Jaser A Zarka, but the cases of meningococcal disease that occurred there could not be classified as an outbreak according to the definition of an outbreak described above.

The identity and characteristics of all isolates were known to only one of the researchers until the genotyping analysis and DHPLC were completed. No data regarding the carrier states of the populations could be obtained.

DNA was extracted from meningococcal cells by the Isoquick DNA extraction procedure (Orca Research), as described previously (6). Amplification and sequencing of the genes evaluated by MLST were performed as described previously (12), with the addition of primers for amplification of the fumc gene (primer fumC-A1 [5′-CAC CGA ACA CGA CAC GAT GG-3′] and primer fumC-A2 [5′-ACG ACC AGT TCG TCA AAC TC-3′]) and sequencing of the fumc gene (primer fumC-S1 [5′-TCG GCA CGG GTT TGA ACA GC-3′] and primer fumC-S2 [5′-CAA CGG CGG TTT CGC GCA AC-3′]). Isolates were assigned to sequence type (STs) as described on the MLST website (http://mlst.zoo.ox.ac.uk/), and the membership of STs to lineages was established with the BURST computer program.

Mutation analysis was performed on a partially inert analysis system from Transgenomic WAVE (Transgenomics Inc., Omaha, Nebr.). The PCR products were denatured at 95°C for 5 min and cooled to 65°C at a temperature gradient of 1°C/min. The samples were kept at 4°C until 5 μl was applied to a preheated C₁₈ reversed-phase column based on nonporous poly(styrene-divinyl benzene) particles (DNA-Sep Cartridge [catalog no. 450181; Transgenomics Inc.]). DNA was eluted within a linear acetonitrile gradient consisting of buffer A (0.1 M triethylammonium acetate [TEAA; catalog no. SP5890; Transgenomics Inc.])-buffer B (0.1 M TEAA, 25% acetonitrile [catalog no. 700001; Transgenomics Inc.]). The temperature at which heteroduplex detection occurred was deduced from Transgenomic software (Wavemaker, version 4.0) and the Stanford DHPLC melting program (http://insertions.stanford.edu/melt.htm), which analyzes the melting profile of the specific DNA fragment. For the identification of alterations in bacterial DNA by DHPLC, the PCR products were analyzed on a 2% agarose gel, and with equal constriction, 250 ng of each of the PCR products from the seven genes was mixed with 250 ng of the PCR product of the same gene from one clone of a serogroup C isolate. The sequences of all seven MLST genes from this clone were known and served as the consensus standards. The mixtures were then denatured and analyzed by DHPLC, as described above. This procedure was repeated for the nine isolates from the outbreak and from an additional 30 control cases to ensure the reliability of the results.

Rep-PCR.

one set of genotypes was generated with two 14-bp oligomers, oligomers Ngrep1R and Ngrep2 (24). A second set of genotypes was generated with a single 22-bp oligomer, oligomer BOXA1R (9). The PCR amplification and the primers used for rep-PCR with N. meningitidis are described elsewhere (28). Dendrograms and similarity coefficients have been analyzed and determined with software from Bacterial BarCodes Inc. (Houston, Tex.).

RESULTS

Over the 40-month period, 16 cases of invasive clinical meningococcal disease (ICMD) were identified in 15 patients (10 males and 5 females) in a population of 25,000 (Table 1). Ten and six cases of ICMD were reported in the pre- and postvaccination periods, respectively, for incidences of 22.8 and 15.1 per 100,000 population, respectively. The mean age at the onset of ICMD was 7.2 years (range, 0.8 to 18 years). One patient was found to be C7 deficient and suffered from two episodes of ICMD over this time period (cases 8 and 12). None of the patients were close relatives. Three cases of ICMD ended in death (cases 5, 15, and 16) (Table 1). Bacterial cultures were positive for 9 of 16 cases of ICMD. Five isolates from the prevaccination period were serogroup C, and four isolates from the postvaccination period were serogroup B. The rate of vaccination coverage was 97%, and all of the patients with ICMD in the postvaccination period had been vaccinated, according to Ministry of Health data.

TABLE 1.

Epidemiological and bacteriological features of the Tamra outbreak

Case no.	Age (yr)	Sexa	Outcome	Date of infection (mo/yr)	Serogroup	ST	ET	Rep-PCR type
1	2	F	Alive	6/96	C	11	37	1
2	0.8	F	Alive	6/96
3	14	M	Alive	2/97
4	6	M	Alive	3/97	C	11	37	1
5	5	M	Death	3/97	C	11	37	1
6	18	M	Alive	4/97	C	11	37	1
7	10	M	Alive	9/97
8	15	M	Alive	12/97
9	2	F	Alive	1/98
10	9	F	Alive	1/98	C	11	37	1
11	4	M	Alive	4/98	B	32	5	2
12	16	M	Alive	12/98	B	Newb	Undetermined	3
13	5	M	Alive	3/99	B	32	5	2
14	1	F	Alive	3/99
15	7	F	Death	8/99	B	32	5	2
16	0.8	M	Death	9/99

^aF, female; M, male.

^bNew set of alleles in some of the genes evaluated by MLST.

MLST.

All five serogroup C isolates from the prevaccination period were shown by MLST analysis to be indistinguishable and to belong to the ET-37 complex. All isolates were ST-11, which is characteristic of ET-37 complex isolates (5). Three of the four serogroup B isolates from the postvaccination period were also indistinguishable by MLST analysis and belonged to the ET-5 complex; they were ST-32, a prototype isolate of the ET-5 complex (2). The only strain from the postvaccination period that differed from the ET-5 complex was the one isolated from the complement-deficient patient (case 12). The allelic profile of this strain has not previously been described in the MLST database.

DHPLC analysis of the nine isolates from the village of Tamra and the control isolates showed that the discriminatory ability of the method by use of heteroduplexes was equivalent to that of DNA sequence determination by MLST (Fig. 1). DHPLC accurately discriminated the control outbreak strains and the control strains from patients with sporadic cases. Genotyping of 39 clones at the seven loci by MLST revealed a 100% correlation between the results of direct sequencing and those of MLST-DHPLC (Table 2). By MLST-DHPLC analysis, the type B strains from the Dir el Asad outbreak and from Jaser A Zarka were classified as ST-32 and ET-5. The type B strains from the Netzarim outbreak and the patients with sporadic cases of type B infection were classified as new ST types; they all differed from each other except for that from one case, whose ST type was identical to that from index case 12 from Tamra. The sporadic type C isolates from different areas throughout the country were all classified as ST-11 and ET-37 by MLST-DHPLC.

FIG. 1.

DHPLC analysis of the pgm genes of nine clones from a meningococcal outbreak in Israel. Elution profiles associated with the PCR amplicons containing the pgm gene (493 bp) were run with a buffer B gradient from 50 to 69% for 5 min at 63°C. Clones 1, 4, 5, 6, and 10 are serogroup C (ST-11, ET-37); clones 11, 13, and 15 (ST-32, ET-5) and clone 12 (new ST, undetermined ET) are serogroup B.

TABLE 2.

Genotyping of N. meningitides clones by direct sequencing and the MLST-DHPLC method^a

Locus fragment	No. of clones with different genotypes by direct sequencingb	No. of clones with different genotypes by MLST-DHPLCc	No. of isolates of the following genotype by sequencing/MLST-DHPLC
			Serogroup Cd	Serogroup Be	Serogroup Wf
fumC	11	11	1/1	9/9	1/1
pgm	9	9	1/1	7/7	1/1
gdh	9	9	1/1	7/7	1/1
pdhc	8	8	1/1g	7/7	1/1
adk	7	7	2/2	5/5	1/1h
aroE	19	19	1/1	17/17	1/1
abc	19	19	1/1	17/17	1/1

^aThe 39 clones were screened for seven of the loci evaluated by MLST. For each locus, the number of different genotypes per among isolates of serogroup C, B, or W are indicated.

^bNumber of clones with variations in sequences.

^cNumber of clones with differences in chromatogram profiles.

^dOf 8 strains tested.

^eOf 29 strains tested.

^fOf two strains tested.

^gThe sequence of one of the serogroup B genotype isolates was identical to the serogroup C sequence.

^hThe sequence of one of the serogroup B genotype isolates was identical to the serogroup W sequence.

Rep-PCR with the Ngrep-based primers and rep-PCR with the BOXA1R primer generated 11 genotype patterns among the 39 isolates in each case. All five isolates from the prevaccination period had identical genotypes by rep-PCR with either the Ngrep-based primers or the BOXA1R primer (Table 1). All three fragments from serogroup B isolates of ST-32 and ET-5 were also indistinguishable by use of the Ngrep-based primers or the BOXA1R primer (Table 1; Fig. 2). The additional serogroup B isolate from the C7-deficient patient had a different genotype (Table 1; Fig. 2). A dendrogram obtained from the analysis with the BOXA1R primer is shown in Fig. 2.

FIG. 2.

(A) Genotypes of N. meningitidis isolates determined by rep-PCR with a single 22-bp oligomer, BOXA1R. The PCR amplification and the primers used for rep-PCR are described elsewhere (6). Lanes 1, 4, 5, 6, and 10, serogroup C (ST-11, ET-37) isolates; lane 12 (new ST, undetermined ET) and lanes 11, 13, and 15 (ST-32, ET-5), serogroup B isolates; lane M, 1-kb DNA size marker. (B) Computer-generated dendrogram of partial rep-PCR analysis of the nine clones of N. meningitidis from the outbreak in Tamra with the BOXA1R primer. The cutoff is 85%; isolates with values over 85% are indistinguishable.

DISCUSSION

Molecular typing can enhance our understanding of the epidemiology of meningococcal infections and improve our ability to design novel means of prophylaxis and treatment (1). The serial cases of ICMD reported on here occurred over a 3-year period in a small village; therefore, we chose to use MLST for molecular typing and for the specific evaluation of slowly accumulating changes. Although MLST has been considered the “gold standard” for the typing of meningococci, it is still rather time-consuming and expensive (29). We used DHPLC in order to simplify and expedite our ability to compare the PCR products from different bacteria obtained by MLST. Since the advantages of DHPLC are accentuated when a sample with a diploid genome generates heteroduplexes due to heterozygous mutations, DHPLC had not previously been used to analyze haploid organisms. In order to confirm the efficacy of DHPLC with DNA fragments in a haploid state, we sequenced the seven housekeeping genes in the bacteria by MLST and then compared them. By using MLST of DNA in the native haploid state, the PCR products in the DHPLC showed no differences that had been noted previously by comparison of the sequences. We attributed the lack of differences to the formation of only homoduplexes, which cannot be discriminated well by DHPLC. In order to overcome this obstacle, we mixed the MLST PCR products of each gene with the MLST PCR product of a specific fixed bacterium that was randomly chosen from the outbreak. This mixture enabled us to create heteroduplexes that were easily identified by DHPLC and totally consistent with the sequencing results (Fig. 1). Use of DHPLC to compare the sequences of the gdh gene fragments among the strains could have reduced the number of sequencing reactions required for the total number of samples (n = 39) to only nine (Table 2) since all fragments with the same chromatogram patterns on DHPLC did not need to be sequenced, thus representing savings in both time and cost.

Modern molecular typing methods with high discriminatory powers are being widely pursued by microbiologists. Such methods have several advantages, including improvements in the precision, portability, and reproducibility of the data obtained. They also enable evolutionary and population studies to be performed with data collected for routine epidemiological purposes (1). We hereby recommend the use of DHPLC to augment molecular typing methods for haploid organisms, whose DNA is converted into a diploid-like state for use in DHPLC.

Repetitive DNA sequences appeared to be universally present in eubacteria and have been applied to fingerprinting of the N. meningitidis genome (11, 24, 28). Rep-PCR analysis provided a degree of discrimination comparable to that of MLST among the meningococcal isolates collected in our study (Fig. 2). These results are consistent with those of a previously described comparison between rep-PCR and multilocus enzyme electrophoresis (24). These results highlight the simplicity and rapidity of rep-PCR as a useful method for real-time analysis of apparent meningococcal outbreaks. Present development efforts are combining rep-PCR with DHPLC to reducing the running time of the typing procedure (unpublished observation).

The molecular typing enabled us to make some important observations regarding the nature of the epidemic. The local authorities declared the cluster of meningococcal cases reported in the village of Tamra as being an outbreak according to epidemiological definitions. The prevaccination period was characterized by the isolation of type C menigococci alone. MLST defined all five isolates as ST-11, the prototype of ET-37 complex strains (29). The clonality of the strains is characteristic of an outbreak (2); these strains had been responsible for major epidemics in the past (2). However, the meningococcal strains cultured in the postvaccination period were all of serogroup B. Three of the four bacteria were found to be ST-32, the prototype of the ET-5 complex, and the ST of one strain was unknown. The serogroup B bacteria (ST-32) attributed to the continuity of the outbreak were part of the ET-5 complex, whose members make up a prominent clonal complex widely distributed throughout the world (2) and whose members are common pathogens in meningococcal outbreaks worldwide. In our study, hypervirulent bacteria of this complex were also isolated form individuals involved in another meningococcal outbreak in Dir el Asad and from patients with serial cases in the village of Jaser A Zarka.

Jackson et al. (8) previously reported the control of serogroup C meningococcal outbreaks using vaccination campaigns for the affected population. Control of the outbreaks is usually achieved within several weeks (8). In the outbreak reported on here, the incidences of ICMD in the pre- and postvaccination periods were 22.8 and 15.1 per 100,000 population, respectively, even though all affected patients had been vaccinated. Apparently, the incidence of ICMD disease was not controlled by the vaccination campaign. However, the bacteriological survey did prove that the serogroup C outbreak was delaminated, and a new cluster of invasive meningococcal serogroup B diseases emerged. In some epidemics, simultaneous or closely linked meningococcal outbreaks due to different serogroups occurred in the same population (10, 17, 19), but this phenomenon has never been described as a possible outcome of a vaccination campaign against N. meningitidis. However, it should be noted that only 5 of 10 bacteria were cultured in the prevaccination period. The other five N. meningitidis isolates from patients with ICMD could have been serogroup B, but this is unlikely since we found that the first and last bacteria from the prevaccination period were both serogroup C and ST-11 (Table 1). Since no other changes in the environmental settings other than vaccination occurred, it is reasonable to assume that all the other strains from the prevaccination period were of the same ST.

What could have been the reasons for the continuity of the outbreak and the serogroup change in the Tamra epidemic described in this study? Underlying immune defects are unlikely, since no known predisposing immunological deficiencies were found other than a complete lack of C7 in a single patient. It should be emphasized, however, that the patients were not evaluated for mannose deficiency. The targeted village of Tamra, located in the lower Galilee region, is characterized by high-density housing and the lack of an internal sewage system; these conditions are common risk factors for meningococcal disease (17).

The experience with the use of the meningococcal serogroup C conjugate vaccine in the United Kingdom has recently been reported on (16), and in that experience an increase in the incidence of serogroup B disease was shown. Concern was raised regarding the possibility of a capsular switch as the result of selection pressure by the vaccine on serogroup C strains (16, 21). It should be noted that both ST-32 and ST-11 isolates display either serogroup B or C on their capsules, implying their potential ability for capsular switching (5). Our study showed no evidence of capsular switching, since the allelic profiles of the housekeeping genes of serogroup B and C isolates differed. However, it is possible that in the specific settings of this outbreak, vaccination allowed the emergence of strains of the hypervirulent ET-5 complex due to selection forces. The complex interactions between the environments, the presence of the hypervirulent ET-5 strain in the population, and vaccination with the quadrivalent vaccine might have contributed to the serogroup change and the persistence of the outbreak.

Studies from the United States (18), The Netherlands (7), and Denmark (13) have demonstrated that infections in complement-deficient individuals are frequently caused by organisms with an unusual polysaccharide capsule. Orren et al. (15) found no ET-5 complex isolates among complement-deficient patients. The reason for this unique observation is unknown and could be related to the low rate of carriage of ET-5 isolates (13). We isolated four different serogroup B bacteria from the village of Tamra. The only non-ET-5 bacterium was cultured from a complement-deficient patient. This observation is consistent with those from the previously described surveys (7).

In summary, we have described the bacteriological features of a meningococcal outbreak determined by using molecular genotyping and highlighted the possible advantages of using DHPLC for high-throughput analysis of haploid organisms. Our data suggest a possible unwarranted outcome of a campaign of vaccination with the quadrivalent vaccine for the control of serogroup C meningococcal outbreaks. Similar evaluations of future meningococcal outbreaks by the methods described above are crucial for understanding meningococcal outbreaks and for the treatment and control of meningococcal outbreaks.