Abstract
In 2000, >400 cases of disease caused by Neisseria meningitidis serogroup W135 (MenW135), the largest MenW135 outbreak reported to date, occurred worldwide among Hajj pilgrims and their contacts. To elucidate the origin of the outbreak strains and to investigate their relatedness to major clonal groups, genotypic and phenotypic subtyping was performed on 26 MenW135 outbreakassociated isolates and 50MenW135 isolates collected worldwide from 1970 through 2000. All outbreak-associated isolates were members of a single clone of the hypervirulent electrophoretic type (ET)-37 complex, designated the “(W)ET-37 clone”; 19 additional MenW135 strains were also members of this clone, and the remaining 31 MenW135 strains were clearly distinct. The 2000 MenW135 outbreak was not caused by emergence of a new MenW135 strain but rather by expansion of the (W)ET-37 clone that has been in circulation at least since 1970; the strains most closely related to those causing the 2000 outbreak have been isolated in Algeria,Mali, and The Gambia in the 1990s.
Neisseria meningitidis serogroup W135 (MenW135) is associated with 1%–8% of all cases of sporadic meningococcal disease worldwide [1–6], and the proportion of healthy carriers in the general population ranges from <1% to 10% [7, 8]. MenW135 is generally considered to have low potential to cause invasive disease or outbreaks; only a few clusters, each of <10 cases, have been reported to date [9].
During March 2000, an increase in the number of cases of meningococcal disease in Saudi Arabia was reported that coincided with the Hajj pilgrimage in Saudi Arabia (14–17 March): of 206 cases, 85 (41%) were confirmed to have been caused by MenW135 [10]. Approximately 1.7 million Muslim pilgrims originating from around the world (1.3 million from outside Saudi Arabia) attended the 2000 Hajj, most during the first half of March. As of August 2000, >400 MenW135 cases had been reported in Hajj pilgrims and their close contacts from16 countries (the United Kingdom, Belgium, the United States, France, Morocco, Kuwait, Saudi Arabia, Oman, Indonesia, Singapore, Denmark, Finland, Sweden, Norway, Germany, and The Netherlands), making this the largest recorded outbreak of meningococcal disease caused by MenW135 [11].
Investigation of this outbreak triggered a number of public health and research questions, the main concern being the origin of the strains associated with the outbreak. We used different phenotypic and genotypic approaches to characterize MenW135 isolates obtained from patients who were infected during the outbreak and to determine the relatedness of these isolates, specifically, whether a singleMenW135 clone was involved. We also evaluated new molecular subtyping approaches (multilocus sequence typing [MLST], multilocus DNA fragment typing [MLDF], and sequencing of the 16S rRNA gene) in an outbreak setting, using as a control a collection of sporadic MenW135 isolates of diverse temporal (1970–2000) and geographic (13 countries) representation.
Materials and Methods
Bacterial strains. A total of 76 strains were characterized in this study. These included 26 strains epidemiologically linked to the Hajj pilgrimage of 2000 and collected from patients with meningococcal disease during March–April 2000 (outbreak associated). The strains were isolated in Saudi Arabia (13 strains), France (4 strains), Singapore (2 strains), Finland (2 strains), and the United States (5 strains). The remaining 50 MenW135 strains were from 13 countries and were isolated from 1970 through 2000 (table 1); these were selected to provide geographic and temporal diversity.
Origin, designation, and results of molecular subtyping of 76 Neisseria meningitidis strains isolated worldwide, 1979–2000.
All strains were characterized using serogrouping, serotyping, serosubtyping, PorA variable region (VR) typing,multilocus enzyme electrophoresis (MEE), pulsed-field gel electrophoresis (PFGE), MLST, and sequencing of the gene encoding 16S rRNA. Fifty-five strains were also characterized using MLDF. In defining a clone, we followed the recommendations of ørskov and ørskov: “the word clone will be used to denote bacterial cultures isolated independently from different sources, in different locations, and perhaps at different times, but showing so many identical phenotypic and genotypic traits that the most likely explanation for this identity is a common origin” [25, p. 346].
Serologic typing and porA sequencing. Serogrouping, serotyping, and serosubtyping were carried out as described elsewhere [12]. PorA VR typing was done by sequencing of VR1 and VR2 of the porA genes [13].
MLST. MLST was performed as described byMaiden et al. [14], by sequencing of 7 housekeeping genes. Primers, determination of sequence alleles, and designation of sequence types are described on the MLST Web site (http://neisseria.org/nm/typing/mlst).
MEE. MEE typing, using 24 metabolic enzymes, was performed as reported elsewhere [15]. Numbers were assigned to enzyme alleles on the basis of enzyme mobilities, and each unique set of alleles was defined as an electrophoretic type (ET). Genetic distances among ETs were calculated as a matrix of pairwise weighted distance coefficients [16]. In this way, differences at less variable loci were given greater weight than those at highly variable loci. Dendrograms were generated by the unweighted pair-group method with arithmetic mean [17], using an SAS Institute program as described by Jacobs [18].
PFGE. PFGE was performed, as described elsewhere, for N. meningitidis [19]. PFGE patterns were designated using the organism, enzyme, and pattern number scheme that is recommended by the Centers for Disease Control and Prevention PFGE committee and used by PulseNet (http://www.cdc.gov/ncidod/dbmd/pulsenet/pulsenet.htm). The complete pattern designations are listed in table 1, but only the numeric pattern portion will be used in the text of this article (e.g., “40” instead of “H46N06.0040”).
16S rRNA gene sequencing. The 16S rRNA genes were amplified by using primers 8F and 1492R [20]. A polymerase chain reaction (PCR) product of ~1.5 kbp was sequenced using the 8F, 1492R, and 8 primers (primers 357, 530, 790, and 981 in forward and reverse orientation), as described elsewhere [20]. An inner fragment of 1471 bp (corresponding to positions 30–1500 of the 16S rRNA of Neisseria gonorrhoeae (Genbank) accession no. X07714) was used for analysis and comparison done with the Wisconsin Package, version 10.1 (Genetics Computer Group).
MLDF. for MLDF, suspensions of bacteria in distilled water (106 cfu/mL) were subjected to 1 freeze-thaw cycle, boiled for 3 min, and then centrifuged for 5 min at 10,000 g. Two microliters of the supernatant were used in each PCR. Three genes, pilA, pilD, and crgA, were amplified by PCR [21]. The primer identifier, primer, gene amplified, (Genbank) accession number, and reference for each target were as follows: for A412 and B127, caatccagcagtcggtccaca and gttgtcggtaacgacgggcag, pilA, X13965, and [21]; for pilD3 and pilD4, tgccgcacagatccggcgcggat and tctcaccggatgggtcagcca, pilD, U32-588, and [22]; and for 98–4 and 98–11, cgttcagccgtgcgcgagagcttggcatgg and gaattatccacgagagattgtttccc, crgA, AF190471, and [23]. PCR was performed as described elsewhere [24]. PCR products corresponding to pilA (1.8 kbp) and pilD (0.9 kbp) genes were then subjected to digestion by each of 3 restriction enzymes: AluI, HpaII, and TaqI. The crgA amplicons (0.85 kbp) were subjected to digestion by each of 4 enzymes: AluI, HpaII, HaeIII, and TaqI. An arbitrary number was assigned to each restriction endonuclease pattern, and an allele was defined by using 3 numbers (4 for crgA) corresponding to the patterns.
Results
When genotypic and phenotypic characterizations were applied to the strains included in this study, 2 major groups were identified. Group 1 was very homogenous and was composed of strains identical to each other in at least 8 of 12 markers assayed. This group included all 26 outbreak-associated isolates and 19 other MenW135 strains. All 45 strains were of the serologic type W135:2a:P1.5,2, except for strainM7165 (isolated in TheGambia in 1995),which was of serologic typeW135:NT:P1.5,2. The other 31 strains formed group 2 and were very heterogeneous. No strain in group 2 showed more than 3 markers that were identical to those in group 1 strains (table 1).
Sequencing of the porA VR1 and VR2 confirmed that group 1 strainswere all identical to the prototype P1.5,2 PorAtype, except for strain M7292 (from South Africa in 1996), which had a variant (VR1 5-1). Strains from group 2 were confirmed to be very heterogeneous by porA VR typing. None of the group 2 strains had the prototype P1.5,2 porA type.
Molecular characterization and subtyping. All strains were analyzed using MLST. With a single exception, all strains in group 1 were sequence type (ST)-11, which is associated with strains of the ET-37 complex (http://neisseria.mlst.net). None of the group 2 strains that we tested was ST-11, and none of the STs found in group 2 is known to be part of the ET-37 complex.
Additionalmolecular characterization of group 1 strains showed that they were all members of a single clone, which is referred to as the “(W)ET-37 clone.” All of the strains in this clone were shown to belong to the ET-37 complex on MEE and MLST analysis; however, this grouping was based not only on these 2 methods but on combined analysis of all 12 molecular markers. Even though a number of methods used in this study have been widely accepted as useful tools for molecular analysis, the designation “(W)ET-37 clone” was chosen because no other classification or grouping provides as wide recognition and association with outbreak potential as does the term “ET- 37 complex.” This clone was composed of all 26 outbreakassociated strains and 19 isolates from cases of sporadic disease in different parts of the world collected during a 30-year period (sporadic/reference isolates of the [W]ET-37 clone): 18 of the sporadic/reference strains that were analyzed by MLST had STs identical to those of the outbreak-associated strains, and 1 of the sporadic/reference strains (M7291) differed in a single allele (table 1).
MEE analysis of the 45 strains in group 1 showed that 36 were ET-27 (figure 1). The remaining 9 strains had 5 closely related ETs that differed from ET-27 in 1–4 enzyme alleles. All of these ETs clustered within the ET-37 complex, with genetic distances <0.22 (table 1 and figure 1). The ratio of strains to ETs for group 1 was 7.5:1, which indicates a low degree of diversity. The 31 strains in group 2 had 26 ETs that differed from ET-27 by 4–12 enzyme alleles (average, 9.45 enzyme alleles; differences were usually seen in enzymes other than those in which differences were seen in group 1 strains). None of these strains was closely related to ET-27 or clustered within the ET-37 complex. The ratio of strains to ETs in this group was 1.2:1, which indicates a very high degree of diversity.
Dendrogram showing genetic relationships of electrophoretic types (ETs) of Neisseria meningitidis isolates of serogroups A, B, C, Y, and W135. From one to several hundred isolates of N. meningitidis isolates are represented by each ET. Arrow points to ET-27.
PFGE analysis showed that a single PFGE type (pattern 40) was identified in all 26 outbreak-associated isolates, as well as in 7 of the sporadic/reference group 1 isolates. Seven additional PFGE patterns, which had >85% similarity (differing from pattern 40 by only 1–3 bands), were identified among group 1 isolates (figure 2). Pattern 42 was shared by 4 isolates, pattern 58 was seen in 3 of the isolates, and patterns 41, 43, 44, 67, and 68 were each seen in 1 isolate (table 1). In contrast, among the 31 group 2 strains, 22 PFGE patterns were seen that had <65% similarity to the cluster containing pattern 40. The PFGE patterns of the group 2 strains had 95% to <60% similarity with each other.
Pulsed-field gel electrophoresis (PFGE) patterns for Neisseria meningitidis W135 isolates. Strains shown are representative for each group. The group 1 patterns are 40–44, 58, 67, and 68.
The 16S fragments (1471 bp) from all 76 strains were sequenced and compared. The differences identified were single-base changes at 18 positions. No gaps were present. The substitutions were distributed along the amplified fragment. At 2 of these positions (in ST-31), apparent mixtures of 2 bases were present. N. meningitidis has 4 rRNA operons [26], and the amplified PCR product could have 1–4 different nucleotides at any single position, depending on differences in the 4 operons. The range of differences among the 16S STs was from1 base (0.07%) to 13 bases (0.91%). A single 16S sequence (ST-31) was identified among the 26 outbreak- associated isolates and a single group 1 sporadic isolate (M7089). Among the 18 remaining group 1 isolates, 16S ST-13 was seen in 15 isolates, and ST-14 was seen in 3 isolates. ST-13 and -14 differ from each other at a single base (at position 1000), and they differ from the ST-31 sequence by 3 nt (ST-13) or 4 nt (ST-14). Among group 2 isolates, twelve 16S STs were identified in 1–7 strains each. None of the strains in group 2 had sequences characteristic of group 1 strains (table 1 and figure 3).
Dendrogram showing the relationships between 15 different 16S rRNA types that were identified among 76 Neisseria meningitidis serogroup W135 strains.
Using MLDF analysis, all 41 assayed isolates in group 1 had identical subtypes in 3 target genes: pilA4, pilD1, and crgA1a (table 1). In one-half of the 14 group 2 strains that were assayed by MLDF, all 3 genes were different from those seen in group 1 isolates, and the remaining 7 strains had either 1 gene (4 strains) or 2 genes (3 strains) identical to those of the group 1 strains (table 1).
Comparison of typing methods. Molecular characterization confirmed that MenW135 isolates in this study belonged to 2 distinct genetic groups. Group 1 contained a total of 45 isolates and included all 26 outbreak-associated isolates. Group 1 isolates were shown to have identical molecular markers when they were assayed by serosubtyping (P1.5,2), MLST (ST-11), and MLDF ( pilA4, pilD1, and crgA1a). In spite of this homogeneity, 2 clusters of very closely related strains were apparent within group 1.
The first cluster (designated “2000MenW135”) contained the 26 outbreak-associated isolates and a single isolate (M7089) with no established connection to the 2000 Hajj. For all molecular subtyping methods, a single subtype was identified: serogroup, W135; serotype, 2a; serosubtype, P1.5,2; VR1 type, 5; VR2 type, 2; PFGE type, 40; MEE type, 27; 16S ST, 31; and MLDF markers, pilA4, pilD1, and crgA1a. The second cluster within group 1 consisted of 19 sporadic/reference isolates collected worldwide from 1970 through 2000 from individuals with MenW135 disease or from MenW135 carriers. Overall, as is shown in table 1, these 19 sporadic isolates were very similar to the 2000MenW135 strains. In fact, 7 of these 19 isolates had a PFGE pattern that was indistinguishable from the pattern typical of 2000MenW135 strains (pattern 40), and the remaining 12 strains had PFGE patterns that had >85% similarity to pattern 40. Similarly, 9 of the sporadic/reference isolates in group 1 had an ET identical to that seen in the 2000MenW135 strains (ET- 27), and the ETs of the remaining 10 isolates showed genetic distance of <0.22. The MLDF types of 15 sporadic/reference group 1 isolates were identical to those seen in the 2000MenW135 strains. None of the sporadic/reference isolates had the 16S rRNA ST 31 that was characteristic of 2000MenW135 isolates, but only 3–4-base differences were identified in the 16S sequences for 2000MenW135 and the sporadic/reference strains in group 1. Overall, analysis of the isolates in group 1 showed not only that 2000MenW135 and sporadic/reference isolates had a large number of identical markers but that the markers that differed were of types that were very closely related to each other.
Group 2 included 31 isolates collected worldwide from 1970 through 2000. Other than serogroup, these isolates were quite different from one another and from group 1 isolates.
Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) in identifying the Hajj outbreak- associated strains were calculated individually for all molecular subtyping methods (table 2). For all methods, sensitivity and NPV were 100%, but 16S sequencing, PFGE, and MEE provided the best specificity (98%, 86%, and 80%, respectively) and PPV (96%, 79%, and 72%, respectively). A combination of PFGE and MEE resulted in a specificity of 90% and a PPV of 84%.
Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of tests used to identify strains associated with the 2000 outbreak of W135 meningococcal disease.
Discussion
The 2000 meningococcal disease outbreak associated with the Hajj pilgrimage to Mecca (Makkah), the largest MenW135 outbreak ever reported, was caused by a single distinct clone (the [W]ET-37 clone) of the ET-37 complex and, in particular, by strains identified here as 2000MenW135. Our data support the initial analyses of a limited number of MenW135 isolates from pilgrims returning to the United States, the United Kingdom, Finland, Singapore, and France [11, 27] and provide evidence related to the origin of this outbreak. Comparison of the outbreak-associated isolates with a large sample of MenW135 isolates from throughout the world collected over the course of 30 years uncovered sporadic disease isolates that had considerable similarity to the outbreak-associated isolates. Those related isolates were recovered in the United States, Canada, Indonesia, Algeria,Mali, The Gambia, Scotland, TheNetherlands, and South Africa as early as 1970.
Two distinct groups of MenW135 were identified. The contrast between the homogeneity of molecular subtypes in group 1 and the high level of diversity in group 2 was demonstrated by multiple approaches. Strains in which at least 8 of 12 markers were identical to the 2000 MenW135 strains made up group 1.
Our data show that MenW135 strains very similar to the 2000MenW135 isolates have been circulating worldwide since at least 1970. Although we cannot exclude the possibility that a new gene was introduced by substitution or insertion into the 2000MenW135 strains, the addition of a new pathogenicity island the size of the smallest observed in the MC-58 genome (5.4 kbp) [26] would have been detected in the PFGE patterns, in which differences as small as 2 kbp have been observed. However, minor changes in genes encoding virulence factors that are important in transmission, colonization, and survival are not likely to have been detected by the approaches used here. Strains M7285 and M7288 (isolated in Scotland and Canada, respectively, in 1970) were found to differ from the 2000MenW135 strains by a single band on PFGE, a single enzyme allele on MEE, and 3 bases on 16S sequencing. These differences in combination could be the result of as few as 5 nucleotide changes and are consistent with the genocloud concept recently used by Zhu [28] to describe the minor variations within the serogroup A strains of subgroup III N. meningitidis.
Several molecular typing methods were used in this study to characterize the (W)ET-37 clone and to evaluate the potential use of new molecular typingmethods not extensively used for outbreak investigations (e.g., MLST, 16S sequencing, and MLDF). In general, all subtyping methods used in our study were very effective in differentiating the (W)ET-37 clone from group 2 strains. Serotyping, serosubtyping, andMLDF were less discriminating; a few identical characteristics seen in the (W)ET-37 clone were also identified in group 2 strains. In contrast, these subtyping methods differed substantially in their ability to differentiate between the 2000MenW135 strains and the related sporadic/ reference strains of the (W)ET-37 clone. 16S sequencing provided the best discrimination, with 98% specificity and 100% sensitivity, for identifying the Hajj outbreak-associated strains. MLST, MLDF, serotyping and serosubtyping, and porA typing had the lowest specificities; the subtypes of most strains of the (W)ET-37 clone were found to be identical by these molecular methods. PFGE and MEE had intermediate specificities.
On the basis of the results of our initial investigation, we adopted an approach that allowed rapid identification of an isolate as the 2000MenW135 strain by using 16S rRNA gene sequencing and PFGE as the first steps in strain characterization. 16S typing was used because it allowed 98% specificity in identification of outbreak-associated strains. PFGE offered significant advantages in rapidity and interlaboratory comparison over MEE andmade it possible to confirm differences. Whether 16S sequencing can be universally applied in future meningococcal disease outbreaks caused by MenW135, as well as in outbreaks caused by other serogroups, remains to be determined. Consequently, the choice of subtyping methods to be used in a particular investigation may depend on the type and extent of individual outbreaks and on each laboratory's requirements and capabilities. This study clearly emphasizes the advantage of the simultaneous use of molecular approaches and epidemiologic data for investigations of meningococcal disease outbreaks.
The origin of the (W)ET-37 clone is of particular interest. In addition to being a member of the ET-37 complex, other molecular and phenotypic markers of this clone are very similar to those of serogroup C strains of the ET-37 complex that are associated with outbreaks and sporadic cases ofmeningococcal disease [29, 30]. This raises the question of whether the (W)ET-37 clone developed from serogroup C by a capsule-switching event [31]. Several examples of closely related N. meningitidis strains that express different serogroup capsules have been documented [32]. In some cases, these strains have been shown to have developed by horizontal transfer of the capsule biosynthesis operon betweenmeningococcal strains [31, 33, 34].Capsular switching from serogroup C to B or to W135 [33] may be selected by mass immunization campaigns in which A+C bivalent polysaccharide vaccine is used.
MenW135 isolates with properties very similar to those of the 2000MenW135 strain have been recovered worldwide since at least 1970 (e.g., a strain isolated from a patient from Indonesia who had just returned from the 1996 Hajj). This suggests that the (W)ET-37 clone has been associated previously with the Hajj and has been disseminated before 2000. It appears that if a capsule switch from C toW135 or vice versa occurred, it must have happened before 2000 and possibly before 1970. Interestingly, MenW135 strains have been isolated from as many as 20% of cases of meningococcal disease in Saudi Arabia every year from 1990 through 2000 (data not shown) [35], but, unfortunately, they have not been retained. The 2000 outbreak appears to be the result of the expansion of the (W)ET-37 clone, members of which have been circulating for a number of years throughout the world. Given that the (W)ET-37 clone has been present worldwide for at least several decades, the Hajj 2000 outbreak was not the result of emergence of a new clone but, rather, an expansion of the already existing one.
Invasive meningococcal disease is caused by a limited number of clonal groups, such as the ET-37 complex, which is primarily associated with serogroup C isolates [29, 36]. Both MEE and MLST, which are molecular typing methods that focus on analysis of genetic coding for metabolic enzymes, identified strains of the (W)ET-37 clone as part of the ET-37 complex. However, in this study, MEE was much more sensitive in differentiating between individual strains within the (W)ET-37 clone (6 ETs were found by MEE, vs. 1 ST by MLST, among group 1 strains). MEE has long been the reference standard for subtyping N. meningitidis, but several features of MLST, such as electronic portability and the unambiguous nature of DNA sequences, are advantageous [14]. Both MEE and MLST data show that characterization of N. meningitidis strains by serogrouping alone will not reveal the pathogenic potential of isolates that belong to recognized virulent clonal groups such as the ET-37 complex. A few MenW135 strains previously have been shown to be members of the ET-37 complex [36], and those findings became even more significant with the occurrence and investigation of the 2000MenW13 outbreak. Using a number of molecular subtyping methods, we demonstrated in the present study that all 8 MenW135strains of the ET-37 complex that were isolated in The Gambia and Mali [37]are indeed members of the (W)ET-37 clone (table 1). Interestingly, ET-24 serogroup C strains, another member of the ET-37 complex, have been the major cause of meningococcal disease outbreaks and sporadic disease in the United States since the beginning of the 1990s [5]. ET-27 strains, which are predominant in the (W)ET-37 clone, differ from ET-24 strains in only 2 of 24 enzymes tested, and no differences in the strains are shown on MLST.
Will the (W)ET-37 clone cause similar outbreaks in future years? The isolation of a strain (strain M7089) belonging to this clone from a patient in the United States who had no known association with the 2000 outbreak but who was associated by time and city (this patient had no known contact with any Hajj pilgrims but lived in a city in which and at a time when Hajjassociated cases were occurring), and of similar W135:2a:P1.5,2 strains in Europe [38] suggest that wide distribution and local transmission have occurred. Caugant et al. [8, 39] reported a 2% MenW135 carriage rate in Norway. More recently, Mac- Lennan et al. [7] reported a 5% MenW135 carriage rate in The Gambia, and, interestingly, 27 of 28 MenW135 isolates studied by that group were members of the ET-37 complex. Further molecular characterization of these strains is needed to assess whether they indeed belong to the (W)ET-37 clone. In Michigan, which is home to the largest Arab population in the United States, no MenW135 carriers were identified in a study conducted in May 2000 on 162 individuals (64 pilgrims who attended the 2000 Hajj, 20 household contacts, and 78 community members; Centers for Disease Control and Prevention, unpublished data), which suggests that MenW135 may have a carriage pattern similar to that of serogroup C strains. Low carriage rates are typical for serogroup C strains of the ET-37 complex, even during serogroup C outbreaks [40]. However, carriage of MenW135 strains was demonstrated in 2 healthy French pilgrims whose family contacts developed meningococcal infections due to 2000MenW135 strains (Institute Pasteur, unpublished data).
The outbreak caused by 2000MenW135 strains, members of the (W)ET-37 clone, has led to a reexamination of current vaccination policies. In 1987, after an outbreak of serogroup A meningococcal disease that occurred in association with the Hajj, Saudi Arabia implemented a meningococcal vaccination requirement for all entering pilgrims [41]. Since the mid-1990s, millions of pilgrims have been vaccinated with bivalent meningococcal A+C polysaccharide vaccine [42, 43]. However, this vaccine provides no protection against disease due to MenW135. A few countries, including the United States, use the quadrivalent meningococcal A+C+W135+Y polysaccharide vaccine that induces serogroup-specific bactericidal antibody, although the clinical protection of this vaccine againstYandW135 serogroups has not been documented [44–46]. One of the pilgrims who became ill on returning from the Hajj to the United States received the quadrivalent vaccine before travel to SaudiArabia. Additional studies are needed to improve understanding of the relationship between vaccination with meningococcal polysaccharides and the carriage or disease potential of MenW135 strains.We predict that outbreaks caused by the (W)ET-37 clone or its relatives will continue to occur.
The 2000 Hajj was the first large worldwide outbreak caused by MenW135 strains. This outbreak clearly was caused by a single clone,members of which have been circulating worldwide at least since 1970 andwhich was associatedwith the Hajj as early as 1996. What changed the interaction between pathogen, host, and environment and caused the MenW135 outbreak in 2000? Available data suggest that little change has occurred in the (W)ET-37 clone since 1970. There are no indications that the conditions at the 2000 Hajj were substantially different from those in previous years. The most dramatic change in the group of Hajj pilgrims has been vaccination with meningococcal A+C polysaccharides since the 1990s. Polysaccharide vaccines are believed to have a minimal long-term effect on carriage, but this is not based on studies of MenW135 or Muslim populations. Furthermore, small, short-term changes in carriage during a period of intense transmission might have contributed to the expansion of this already existing clone. In the Africanmeningitis belt, large serogroup A epidemics may be masking small, low-intensity MenW135 epidemics. Therefore, laboratory-based, populationbased surveillance is crucial for early recognition of infrequently seen meningococcal serogroups and implementation of appropriate vaccination programs.
Coordinated efforts on the global level are needed to monitor the spread of the (W)ET-37 clone and define its virulence, as well as to understand the role of vaccination programs on the dissemination and evolution of the clone. Molecular characterization of N. meningitidis, as demonstrated in this study, is a powerful tool for understanding the natural history of meningococcal disease and formulating recommendations for its control and prevention.
Acknowledgments
We are indebted to the following for collaboration and for supplying strains: Maija Toropainen (Laboratory of Vaccine Immunology, Department of Vaccines, National Public Health Institute, Helsinki), Raymond Lin (consultant, Clinical Microbiology Laboratory, Women's and Children's Hospital, Singapore), Mavis Yeo (Head, Diagnostic Bacteriology, Department of Pathology, Women's and Children's Hospital), F. Mostashari (New York City Department of Health, New York, and Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta), and N. Bendana (California Department of Health, Los Angeles). We also thank Mark Achtman, David Stephens, and Brad Perkins, for suggestions and critical review of the manuscript; Brian Plikaytis, for statistical advice; Gwen Barnett, for help with the figures; and Rana Hajjeh and many others, for useful discussions.
