The introduction of PCV7 has resulted in a dramatic reduction in invasive pneumococcal infections caused by the serotypes targeted by the vaccine (33
). However, the long-term effectiveness of the vaccine is not clearly known, in part due to the unknown extent that replacement disease—an increase in the incidence of non-PCV7 serotype disease—will become a factor. It is possible that PCV7 could have profound effects upon the prevalence and strain composition of individual serotypes that it does not target. The marked increase in the incidence of invasive disease due to serogroups not included in PCV7 (and serotype 19A) in 2002 relative to that in 1999 (P
≤ 0.001) provides ample rationale for continued monitoring of individual serotypes within this set (Table ). It follows that certain individual non-PCV7 serotypes (serotypes 3, 7F, 15B/C, 19A, 22F, 33F, and 38) individually accounted for significantly higher proportions of invasive cases among children <5 years of age in 2002 relative to the proportions in 2001 (Table ). These data are consistent with those from a recent report of the increase in the number of invasive pediatric cases due to serogroups 15 and 33 in the post-PCV7 era (16
). The cumulative data indicate a need for continued monitoring of invasive pneumococci to detect changes in serotype-genotype associations and to determine the effects of the vaccine on the pneumococcal population genetic structure. The focus of this work has therefore been to provide continuing serotype-specific data on invasive strains and to compare those data to our pre-PCV7 data (12
Sharing of the same ST among two or more serotypes is presumably indicative of past horizontal transfer events of capsular type-specific loci. Our data suggest that such capsular switching is a rare event since during the 3 years of this molecular surveillance study and the analysis of over 2,100 isolates we have detected only 11 STs associated with multiple serotypes. This is consistent with the findings of a recent study which found no evidence of capsular switching among 333 carriage isolates recovered from 100 children, including all isolates of different serotypes recovered from the same child (20
). It could be argued that we did not identify more putative instances of capsular switching because only 285 of these isolates were actually submitted to MLST (data not shown). We contend that our strategy of strategic MLST of representative related clusters within serotypes and of isolates involved in all unusual PFGE associations between isolates of divergent serotypes actually allows the targeting of more unusual multilocus sequence type associations than a strategy that uniformly targets a more limited number of isolates for MLST. In only 1 of the 11 instances of a shared ST between isolates of different serotypes (STs 62923F,23A
) was the ST (ST199) associated with both PCV7-PCV7R and PCV7NR serotypes. However, it is noteworthy that ST199 accounted for the majority of serotype 15B/C and 19A isolates during 1999, even before the introduction of PCV7 (12
). The remaining 10 STs were shared only between isolates of PCV7 or PCV7R serotypes. Our data and the global data reveal relatively few instances in which identical or highly related STs are associated with both PCV7-PCV7R and PCV7NR serotypes. These findings are in marked contrast to those presented in a recent report (15
) describing 11 instances of STs shared among multiple serotypes from a sample set of only 252 isolates, in which four STs were shared among PCV7-PCV7R and PCV7NR serotypes and in which three STs accounted for three to four different serotypes that included PCV7-PCV7R serotypes together with PCV7NR serotypes. In general, the data from our study indicate an extensive history of capsular switching between penicillin-nonsusceptible strains restricted to the PCV7 and the PCV7R serotypes but few deduced instances of genetic exchange involving strain pairs that represent both a PCV7-PCV7R serotype and a PCV7NR serotype. CC13
, and CC1296
are examples of clonal complexes documented here and elsewhere that are primarily associated with antibiotic-resistant strains and multiple PCV7 and PCV7R serotypes.
Of interest was the single example in which a single PCV7NR serotype (serotype 15B) PI had an ST (ST162 within CC156
) that was previously reported primarily among strains with PCV7 serotypes (28
). Within CC156
, we found an ST162 serotype 15B isolate recovered from the blood of an infant. ST162 is primarily associated with penicillin-sensitive and intermediately penicillin-resistant strains with PCV7 serotypes 19F, 9V, and 14 (28
). The recent observation of multiply resistant serotype 11A isolate derivatives of the predicted founder strain Spain9V
) within CC156
also supports the need for the continued monitoring of the clonal composition of serogroup 15 (29
). Although such occurrences appear to be rare at present, the constant usage of PCV7 may aid the expansion of such serotype switch variants with implications for the long-term effectiveness of the current vaccine.
The clonal structures of most serotypes characterized appeared to be stable by comparison to the prevaccine clonal structure data. Only 9 of the 83 clonal sets included both PCV7-PCV7R and PCV7NR serotypes, with only 3 of these clonal sets (CC156, CC1292, and CC899) in the post-PCV7 era showing new associations not previously seen in the pre-PCV7 period. Furthermore, three clonal sets which consisted primarily of PCV7NR serotypes prior to the introduction of PCV7 showed an association with PCV7-PCV7R serotypes only after PCV7 introduction (CC62 and CC63) or single isolates prior to PCV7 introduction (CC662). The other examples of such associations with both PCV7-PCV7R and PCV7NR serotypes include CC199, CC460, CC1257, and CC199. It is likely that CC460 was common within both serotype 6A and serotype 10A isolates before and after PCV7 introduction (Fig. ). CC1257 was seen among two serotype 20 isolates (ST1257) in 2001 and in a single serotype 9N isolate (ST1375) in 2002. Since multiple SLVs and DLVs of these STs are documented at mlst.net in association only with serotype 20 isolates, it is likely that CC1257 originated within serotype 20.
Of the nine clonal sets inclusive of both PCV7-PCV7R and PCV7NR serotypes, CC199 is currently the most important. Before and after the introduction of PCV7, CC199 accounted for the majority of both serotype19A and 15B/C isolates (Fig. ). Recent ABCs data indicate that the rate of invasive pneumococcal disease associated with serotype 19A is on the increase (23
). Also, other reports have indicated that serogroup 15 is an important component of serotype replacement occurring in vaccinated children (13
). Serogroup 15 and serotype 19A showed the most pronounced increases in incidence among invasive pediatric isolates in 2002 relative to the incidence in 1999 (Table ); and these increases were associated with a corresponding increase in the incidence of CC199, which constituted 70 to 92% of both the serogroup 15 and the serotype 19A isolate sets in 2001 and 2002 (Fig. ). This trend has continued among the more recent year 2004 isolates analyzed (23
Since ST199 was the most common ST from both serotype 15B/C and serotype 19A isolates recovered from 1999 to 2002, presumably serotype 15B/C and 19A CC199 strains have a recent common ancestry. In contrast to serotype 15B/C isolates, which are penicillin susceptible, the majority of serotype 19A isolates are penicillin nonsusceptible, indicating that the serotype 19A derivative of ST199 might have originated following the transformation of a type 15B/C ST199 strain with serotype 19A-specific cps
locus genes. It appears that PCV7 does not protect against the PCV7-related serotype 19A, which is consistent with the findings in a recent report in which a nine-valent conjugate vaccine protected against carriage by vaccine types and vaccine-related serotype 6A, with a concomitant increase in the rate of carriage of nonvaccine types (6
). Significantly, this vaccine offered reduced protection against vaccine serotype 19F and did not reduce the rate of carriage of serotype 19A strains (6
). These findings are also consistent with the results found for serotypes 19A and 19F in a clinical trial assessing the efficacy of PCV7 against acute otitis media (9
). PCV7 may have created an improved niche within the pediatric population for the proliferation of strains with specific nontargeted serotypes.
The cumulative data from this genotypic surveillance indicate that clonal associations remained relatively stable overall from 1999 to 2002, with PCV7 potentially having a modest, yet important, impact on the clonal composition within individual serotypes during 2001 and 2002. For example, although serotype 19A showed a constant proportion of CC199 isolates during the 3 years of surveillance, we have noted an increase in the rate of antibiotic resistance in 2003 and 2004 among isolates within this serotype (23
), due in part to the continuing presence and increased incidence of clonal complexes, such as CC236, CC1296, and CC81, that first appeared in our serotype 19A surveillance isolates during 2001 and 2002 (Fig. ; Table ).
Among the PCV7NR serotypes, two additional clonal types have accounted for a larger proportion of pneumococcal isolates from invasive pediatric disease cases within the ABCs following PCV7 implementation. CC62 (which a showed a more than twofold increase in both years, with 14 to 20 serotype 33F isolates, respectively), and ST393 (which a showed a more than threefold increases incidence in 2002 relative to that in 1999, with 17 serotype 38 isolates) were observed to be the predominant clonal complexes within their respective serotypes even before PCV7 implementation (Fig. ; Table ). These data correspond to the overall increase in the incidence of disease caused by serotypes 33F and 38 in 2002 relative to the incidence in 1999 (Table ). Extended surveillance will monitor the potential for these two clones to become common invasive pathogens in children.
Although the genetic structure of most serotypes remained relatively stable, the composition among some serotypes (e.g., serotypes 6A, 15B/C, and 19A) appeared to be altered to some degree through the occurrence of additional clonal sets when pre- and postvaccination data were compared (Fig. ). Although clonal sets were assigned for only 72 to 77% of the serotype 6A and 19A isolates from 1999 (Fig. ), we were unable to find matches between unassigned PFGE profiles from year 1999 isolates with PFGE profiles from the 2001 and 2002 data set. Therefore, it is unlikely that new clonal groups within these serotypes that were identified during 2001 and 2002 accounted for a major portion of unassigned year 1999 isolates. ST1292, one of the clonal sets not found within the 1999 serotype 6A isolates, shared only four allelic identities with its closest match at mlst.net at the time of its discovery within the 2001 isolate set. This ST is an SLV of ST1379 subsequently reported from a serotype 6A blood isolate recovered in Sweden in 2000 (28
). It is interesting that within this nine-isolate CC1292 set, a full range of penicillin resistance phenotypes is evident (sensitive, intermediately resistant, and fully resistant), indicating that this strain might be rapidly acquiring resistance.
Although this study has documented the maintenance of established serotype-clonal group associations and has documented new associations in the post-PCV7 era, it has limitations. The findings here are limited to strains obtained immediately after the introduction of PCV7. Pneumococcal populations evolve gradually, warranting long-term surveillance. Also, the study does not account for all pneumococcal isolates collected in the ABCs database, although the majority of the pediatric isolates from each year are accounted for (this is true even for 2002, where PCV7 serotypes were largely omitted). Nonetheless, the data do serve as a continuing reference point for future molecular studies that can be used to assess the impact of the vaccine upon the spectrum of invasive isolates collected from children in the United States. Additionally, targeted analysis of specific isolates from older individuals allows assessment of the clonal types associated with important antibiotic resistance.
Genotyping of surveillance isolates led to a fairly extensive list of corrections of the phenotypic data, primarily due to the retesting of isolates with unusual serotype or resistance associations with STs. A modest degree of serotyping error has been found within other reference laboratories (17
), emphasizing that additional levels of quality control are desirable. We are currently targeting all isolates with serotypes associated with unusual antibiotic resistances or genotypes for the detection of potential phenotyping errors. The pneumococcal MLST database has proved very valuable for screening of serotype associations in this regard.
Recent studies have indicated that it is possible that pneumococcal serotypes, rather than specific clonal features, are the primary predictors of pneumococcal virulence (3
). These studies have relied upon the relative number of isolates of specific clones and serotypes isolated from invasive infections relative to their occurrence during carriage. A contrasting report of a study that used a similar strategy indicated that specific clonal features other than the serotype could be important predictors of virulence (30
). It is possible that this general approach is misleading, since in many geographic locations a specific serotype might be limited to very little clonal variation. It is also possible that insufficient numbers of isolates within specific clonal types were recovered from either invasive infections or carriers, making comparisons difficult. Other potentially confounding aspects include the potential to not detect cocolonizing carriage strains and not detect carriage strains present in the upper respiratory tract in small numbers. One recent report (22
) of a study that used a mouse model clearly showed profound differences in the invasiveness of two strains of various backgrounds that shared the same serotype and showed the similar invasiveness of two different serotypes that shared an otherwise identical genotype. It is hoped that further investigations will shed light on this issue. Continued meaningful integration of genetic identifiers, phenotypic data, disease manifestations, and demographic information will help to provide a better understanding of the epidemiology of this important pathogen.