This study shows that S. pneumoniae
was the leading species-specific cause of childhood meningitis during the last decade in Bangladesh, accounting for 46% of culture-positive cases. Serotype-2 pneumococcus was the leading serotype-specific cause during the surveillance period, accounting for 20% of the culture positive meningitis cases with a very strong predilection for causing meningitis (OR
29.6) compared with bacteremic IPD. Another distinctive feature of this pneumococcal serotype was that it affected younger children compared with all other serotypes and it also consisted of a genetic lineage of highly related STs.
Routine, comprehensive population-based surveillance is not established throughout Bangladesh 
, which prevents determination of the nation-wide incidence of meningitis. Consequently, the full extent and impact of serotype-2 pneumococcal meningitis is difficult to ascertain. The study, however, took advantage of rigorous, ongoing hospital-based surveillance with concomitant careful laboratory investigation of cases. Serotype-2 was also sporadically isolated during the period of passive surveillance from 1993 to 2000, when 15 cases were identified (~ two cases per year) 
. Since the end of 2009, four culture-positive serotype-2 cases, all with meningitis, have been recorded (data not shown). Since we conducted surveillance in several large hospitals and used sophisticated laboratory methods to identify the serotypes, we are confident that serotype-2 pneumococcal meningitis over the observation period was the predominant cause of pneumococcal childhood meningitis. The recent observation of fewer cases may be chance variation or alternatively suggestive of epidemic like behavior typical of serotype-1, which is known to cause epidemics of IPD and specifically meningitis in West Africa 
The geographic range and burden of serotype-2 IPD in Bangladesh and surrounding countries is difficult to determine because active surveillance with serotype testing in regions surrounding Bangladesh is limited or non-existent. The number of cases identified in this study is likely to be an underestimate of the burden, as only a minority of sick children with symptoms of meningo-encephalitis present to hospitals in Bangladesh 
. Also, the very substantial exposure to antibiotics (>50% of meningitis cases) likely reduced the detection of cases. For these reasons, it is possible that a major burden of serotype-2 pneumococcal meningitis could go unrecognized. Intriguingly, a small-scale report of 28 invasive pneumococcal disease patients from Nepal 
included two cases (ranking 3rd
in serotype prevalence) caused by serotype-2, which indicates the existence of this serotype in a neighboring country. Furthermore, a wider spectrum of serotypes could contribute to pneumococcal meningitis, as 93 (43%) of the 217 antigen-positive CSF samples failed to yield a result on serotype-specific PCR that was limited to about a third of the 90 possible serotypes identified.
In the 1930 s, serotype-2 was a major cause of pneumococcal pneumonia but apparently not meningitis. In a large series of 991serotype-2 adult pneumonia cases observed in the US , 46% of which were bacteremic, only 24 cases (2.4%) manifested meningitis 
. Recent reports on 70 years of pneumococcal serotype secular trends from the USA and Denmark showed disappearance in the 1960 s of serotype-2 from reports of IPD 
. In the 1970 s, serotype-2 was still a major cause of pneumonia among South African miners 
, but was not overtly described as having a propensity for causing meningitis. Serotype-2 IPD is now rare in South Africa 
. No recent reports of high-incidence serotype-2 IPD could be found in the published literature.
It is not possible to determine whether Bangladesh is a major persisting reservoir of serotype-2 pneumococci or whether this serotype-2 meningitis-causing lineage is emerging as a new variant. Genetically, the lineage occurring in Bangladesh is distinct from the historic strain D39 isolated in the early 1900 s.The meningitis-causing feature of the serotype-2 pneumococcal lineage reported here is unusual or even unparalleled for a pneumococcus in having such a high odds for causing meningitis. In studies of meningitis from Africa, compared to serotype-1, serotype 2 did not account for a prominent proportion of meningitis cases 
. It raises the question of whether a variant with specific central nervous system tropism has recently evolved within the clonal complex CC74 in Bangladesh which differentiates it from the members of the same lineage observed in West Africa 
. Typically, such a new bacterial variant would undergo clonal expansion yielding indistinguishable isolates. The PGFE and MLST typing results are broadly consistent with such a phenomenon. However, the MLST data indicate there are three closely related STs: ST 74, ST 5083 and ST 5199 (the latter two being unique to Bangladesh) all belonging to the same clonal complex CC 74. These data are commensurate with these three members of CC74 having circulated for sufficient time to accumulate such sequence diversity. The three STs occurred consistently over the observation period, suggesting that evolution of this diversity antedated the sampling period. A better understanding of the possible genetic basis for this meningitis-causing trait awaits further investigation, including whole genomic sequence analysis and comparative genomics, which is ongoing.
The high proportion of meningitis associated with this serotype-2 pneumococcus strain highlights the potential for a serotype that was perceived to have nearly vanished to materialize as a major cause of IPD and pneumococcal meningitis in particular. Owing to its rarity in recent reports of IPD, serotype-2 has not been deemed a likely candidate for inclusion in any currently planned conjugate vaccine. It is unclear whether serotype-2 represented by this clonal complex will remain a dominant cause of meningitis in Bangladesh, spread to neighboring countries or even re-emerge as a major cause of IPD in regions where it had disappeared.
This study confirms the importance of ongoing surveillance before and after vaccine introduction, as well as the potential for substantial shifts in serotype profiles in the absence of vaccine introduction. It also indicates the need for alternate vaccine candidates that provide protection across serotypes such as protein-based vaccine candidates.