Abstract

BackgroundThe relative invasiveness rates (attack rates) of Streptococcus pneumoniae of different capsular serotypes in children are not known. Estimates of capsular serotype invasiveness (designated “invasive odds ratios”) that are based on cross-sectional prevalence carriage data have been published, but these estimates could be biased by variation in the duration of carriage

MethodsThe relative attack rates of invasive pneumococci were measured using national UK surveillance data on invasive pneumococcal disease (IPD) incidence and data on incidence of pneumococcal acquisition from longitudinal studies of nasopharyngeal pneumococcal carriage

ResultsWe found significant differences in capsular serotype–specific attack rates. For example, capsular serotypes 4, 14, 7F, 9V, and 18C were associated with rates of >20 IPD cases/100,000 acquisitions, whereas capsular serotypes 23F, 6A, 19F, 16F, 6B, and 15B/C were associated with <10 IPD cases/100,000 acquisitions. There was an inverse relationship between duration of carriage and attack rate by capsular serotype (P<.0001). Attack rates were significantly correlated with invasive odds ratios (P<.0001)

ConclusionsThe capsular serotype is a major determinant of both pneumococcal duration of carriage and attack rate. Published invasive odds ratios are a reliable and practical method of determining capsular serotype invasiveness and will be valuable for investigating and characterizing emerging capsular serotypes in the context of conjugate vaccination

Streptococcus pneumoniae is a human-restricted bacterial commensal. Most, if not all, people acquire it at some time, and it usually colonizes the nasopharynx of individuals harmlessly. Of the pneumococci acquired by individuals, a few will go on to cause a spectrum of mild and severe diseases. Accumulatively, the diseases caused by pneumococci make it one of the leading causes of infectious diseases in people of all ages, but particularly in young children and the elderly. The principle disease manifestations are pneumonia, bacteremia, meningitis, and otitis media. Worldwide, it is estimated to cause up to 1 million deaths per year in children <5 years of age

Recent publications have evaluated the relationships between strains for their potential to cause disease. A major finding was that strains, whether categorized by sequence type or capsular serotype, vary substantially in invasive potential, as determined by an “invasive odds ratio” [1, 2]. This finding suggests that most of the invasive property of pneumococci is determined by their capsular serotype rather than their genetic background, as determined by multilocus sequencing. Therefore, comparison of the invasive potential of capsular serotypes provides a useful approach for comparing the behavior of pneumococci. This invasive property of pneumococci is independent of time and geography, suggesting a surprising stability of pneumococcal capsular serotypes under natural conditions [2]. This approach to characterizing strains is particularly relevant to assessing any changes in the invasive disease potential of capsular serotypes that emerge after the new immune selective pressure caused by implementation of pneumococcal-conjugate vaccine use. However, the odds ratio calculation was based on the prevalence (or carriage) of a strain/capsular serotype and the frequency of invasive disease isolates. Such an approach assumes a constant duration of carriage. Because it is well known that pneumococcal capsular serotypes vary in their duration of carriage [3, 4], the extent to which this variation in duration of carriage accounted for the differences in the invasive odds ratio was not known

Estimates of “invasiveness,” such as the invasive odds ratio, not only represent a biologically intriguing characteristic of a capsular serotype, they may also be used to make informed decisions regarding future vaccine formulations, and, as such, it is important that they are as precise as possible. Whether invasive odds ratios are reliable indicators of a real difference in invasiveness between capsular serotypes can be determined only if such results are compared with a measure of invasiveness that is independent of duration of carriage. This can be achieved by using pneumococcal acquisition data rather than carriage data in the calculation

The present article presents novel data that demonstrate the varying capacity of capsular serotypes to cause disease once acquired. To our knowledge, it is the only article that presents data of a longitudinal nature on both invasive pneumococcal disease (IPD) and acquisition in the same region over the same time period. It also provides the opportunity to assess the reliability of new approaches that estimate invasive potential by use of cross-sectional data

Methods

Study designTwo sets of data collected in Oxfordshire, United Kingdom, were used to directly measure acquisition and duration of carriage. When considered together, these studies comprise 313 infants <2 years of age, representing 13,460 child-weeks (or 280 child-years), and >3000 pneumococcal isolates. The following 2 studies were undertaken: In study 1, a birth cohort of 100 children was followed for 24 months between 1996 and 1999. Nasopharyngeal (NP) swabs were obtained at 4, 8, 12, 16, 20, 24, 32, 40, 48, 56, 64, 72, 80, 88, and 96 weeks of age. A total of 1353 (90.2%) of 1500 planned visits took place. In study 2 [5], a birth cohort of 213 infants was followed for 24 weeks between 1999 and 2001. NP swabs were conducted at 2, 4, 6, 8, 10, 12, 16, 20, and 24 weeks of age. A total of 1907 (99.5%) of 1917 scheduled visits took place

These longitudinal carriage studies were conducted simultaneously with active surveillance of IPD in Oxfordshire since 1995. National surveillance data from the Public Health Laboratory Service (PHLS; now named the Health Protection Agency) during 1995–1997, as previously reported, was used. Details of these 2 surveillance systems and the data collected from them are outlined in previous publications [6, 7]

DefinitionsInvasive disease cases were defined by the isolation of S. pneumoniae from the blood, cerebrospinal fluid, or other normally sterile sites, as described elsewhere [7]. Only 1 isolate per disease episode was analyzed. Episodes were considered as 2 separate cases if the same capsular serotype was found in another sample taken at least 30 days after the previous culture of a sample

Acquisition was defined as isolation of a new S. pneumoniae capsular serotype not previously detected in the 2 previous visits, as described elsewhere [5]. Carriage was defined as starting halfway between the last negative swab and positive swab. Termination of a carried capsular serotype was defined as loss of a capsular serotype for at least 2 consecutive visits. Time was, therefore, censored at week 18 for carriage study 1 and at week 80 for carriage study 2

Microbiological analysisS. pneumoniae were cultured from the nasopharynx, using a plain cotton-tipped per-nasal pediatric swab (lot no. 00A26; Medical Wire and Equipment), as reported elsewhere [5]. Collection, transport, processing, culturing, and serotyping of the NP samples were conducted in accordance with the World Health Organization protocol [8], except that cotton, rather than calcium alginate, NP swabs were used to enable comparisons between the carriage studies conducted in Oxford, as described elsewhere [5]. NP swabs were obtained by inserting the swab fully into the posterior nasopharynx. The swab was left in place for at least 5 s before removing it slowly. All NP swabs were placed in a transport media (skim milk, glycerol, glucose, and tryptone soya broth), taken to the Department of Microbiology at the John Radcliffe Hospital, and plated within 8 h of collection. Three morphologically distinct colonies were selected for serotyping. Serotyping was performed by use of the Quellung reaction with capsular serotype–specific antiserum (Statens Seruminstitut), as described by Lund and Henrichsen [9]

Analytical methodsThe capsular serotype–specific incidences of IPD and acquisition in children <2 years of age were calculated using the standard formulas: Incidence of IPD was calculated as the number of new invasive disease cases divided by the total population at risk, in child-years. Incidence of acquisition was calculated as the number of new nasopharyngeal acquisitions divided by the total population at risk, in child-years

Duration of carriage is presented as the half-life that each capsular serotype was carried. To allow for false negative swabs, termination of carriage was defined as 2 consecutive negative swabs. Hence, observations were censored at week 18 for carriage study 1 and at week 80 for carriage study 2, thereby, allowing 2 observation points before the end of sampling to define termination of carriage. The duration of carriage for each capsular serotype was estimated using survival analysis with a simple exponential decay model. Because estimates of duration were used to compare capsular serotypes within the study, no adjustment (with the necessary model-dependant assumptions) for interval censoring was made

The incidence data for England and Wales were used, because it had previously been shown that there was no detectable difference between the data from the enhanced Oxford surveillance of IPD and that from England and Wales [7]. For England and Wales, the incidence of invasive disease for children <2 years of age for the study period was calculated using a denominator of 4,455,500 population-years, as determined from national data as described elsewhere [7]. For the carriage studies, the incidence of acquisition was calculated using a denominator of 280.4167 population-years. The capsular serotype–specific attack rate was defined as the ratio of the incidence of disease to the incidence of acquisition and was calculated to determine the number of cases per 100,000 acquisitions (table 1). Confidence intervals for the attack rates were determined using Monte Carlo simulations (10,000 runs) of the ratio of invasive episodes to acquisitions, assuming a Poisson distribution

Table 1

Invasive, acquisition, and carriage properties of pneumococcal capsular serotypes

Table 1

Invasive, acquisition, and carriage properties of pneumococcal capsular serotypes

Invasive odds ratios and duration of carriage were correlated to attack rates. Invasive odds ratios were taken from a meta-analysis of 7 different data sets of invasive and carried pneumococcal capsular serotypes recovered from children [2]. Attack rates were compared with log-transformed invasive odds ratios and with duration of carriage by use of regression analysis. All statistical analyses were performed using Stata (version 7.0; StataCorp)

Results

Capsular serotype distributionThis study identified 36 different capsular serotypes carried in the nasopharynx. By 6 months of age, 54% of infants had acquired pneumococci at least once. By 2 years of age, this percentage had increased to 97% of all children. Of all the children who acquired pneumococci, 48% acquired ⩾1 capsular serotype, and 13% had ⩾1 capsular serotype isolated on at least 1 swab

The incidence of IPD, the incidence of acquisition, attack rate, and duration of carriage are presented in table 1. The most commonly isolated capsular serotypes from invasive disease surveillance (i.e., those with the highest incidence) were capsular serotypes 14 (30.6%), 19F (8.6%), 6B (8.1%), and 23F (2%). The highest acquisition rates were observed for capsular serotypes 6B, 19F, 23F, and 14 (between 10 and 32/100 child-years), whereas the lowest acquisition rates were observed for capsular serotypes 9V, 10A, 11A, and 21 (between 3 and 5/100 child-years). There was variation in the duration of carriage for those with >10 acquisitions; 15C had the shortest duration of carriage (5.9 weeks), and 6B had the longest (19.9 weeks)

Incidences of acquisition and invasive diseaseThere was a wide range in the incidence of acquisition of pneumococci and the incidence of invasive disease between different capsular serotypes (figure 1). For example, capsular serotype 14 was responsible for significantly fewer acquisitions than capsular serotypes 19F and 6B but for a significantly higher incidence of disease

Figure 1

Correlation between the incidence of pneumococcal acquisition and the incidence of invasive pneumococcal disease (IPD) for each capsular serotype. Data on capsular serotypes 1, 5, and 9A are estimates because there was no acquisition of these capsular serotypes in the carriage data sets. Confidence intervals are indicated by error bars

Figure 1

Correlation between the incidence of pneumococcal acquisition and the incidence of invasive pneumococcal disease (IPD) for each capsular serotype. Data on capsular serotypes 1, 5, and 9A are estimates because there was no acquisition of these capsular serotypes in the carriage data sets. Confidence intervals are indicated by error bars

Attack ratesThe attack rates were calculated for each capsular serotype and are shown in table 1. A wide range of attack rates were found. For example, capsular serotypes 1, 5, 9A, 4, 14, 12F, 7F, 8, 9V, 18C, and 19A are associated with high attack rates, with 20–80 cases of IPD/100,000 acquisitions of the respective capsular serotype, whereas capsular serotypes 38, 3, 23F, 6A, 19F, 16F, 6B, 9N, 20, 35F, 10A, 15B/C, 11A, 15A, 17F, 33F, 18B, 13, 23A, 22F, 21, 31, 23B, and 37 are associated with <10 IPD cases/100,000 acquisitions of the respective capsular serotype. Data for capsular serotypes 18B, 13, 23B, 37, 31, 17F, and 15A are unreliable because of small numbers (<5 acquisitions and no reported case of invasive disease). Those for capsular serotypes 1, 5, and 9A are estimates because there were no detected acquisitions. These data highlight 4 capsular serotypes (7F, 8, 12F, and 19A) and that would be considered to have high attack rates (⩾20 cases/100,000 acquisitions) and that are not included in the currently marketed heptavalent vaccine (which includes the following capsular antigens: 4, 6B, 9V, 14, 18C, 19F, and 23F)

Relationship between duration of carriage and disease There was an inverse relationship between the attack rate of a capsular serotype and its duration of carriage (P<.0001) (figure 2). Those capsular serotypes that are able to survive in the nasopharynx of children for long periods are significantly more likely to be those capsular serotypes with low attack rates (e.g., capsular serotypes 16F, 21, 33F, 15A, 6B, and 35F). Conversely, those capsular serotypes with short carriage duration are associated with higher attack rates (e.g., capsular serotypes 14, 12F, 7F, 9V, and 18C)

Figure 2

Correlation between the attack rate and duration of carriage (half-life in weeks) for each capsular serotype. Capsular serotypes 1, 5, and 9A are crude estimates. Because these capsular serotypes were isolated only from patients with invasive pneumococcal disease (IPD), but not from the nasopharynx, the duration of carriage was assumed to be <4 weeks and the attack rate was estimated to be >60 cases/100,000. A regression was conducted on all capsular serotypes apart from capsular serotypes 1, 5, and 9A. The relationship between the 2 variables is linear and statistically significant (P<.0001). Capsular serotypes in the heptavalent vaccine are indicated in bold type

Figure 2

Correlation between the attack rate and duration of carriage (half-life in weeks) for each capsular serotype. Capsular serotypes 1, 5, and 9A are crude estimates. Because these capsular serotypes were isolated only from patients with invasive pneumococcal disease (IPD), but not from the nasopharynx, the duration of carriage was assumed to be <4 weeks and the attack rate was estimated to be >60 cases/100,000. A regression was conducted on all capsular serotypes apart from capsular serotypes 1, 5, and 9A. The relationship between the 2 variables is linear and statistically significant (P<.0001). Capsular serotypes in the heptavalent vaccine are indicated in bold type

Relationship between invasive odds ratio and attack rate Invasive odds ratios were significantly correlated with the attack rate (P<.0001) (figure 3). Because capsular serotype 12F, which had a high attack rate (38 cases/100,000 acquisitions), was not included in the reports of pneumococcal invasive odds ratios [1, 2], this capsular serotype was excluded from the analysis

Figure 3

Relationship between the invasive odds ratio derived from cross-sectional data and the attack rate measured directly from longitudinal data. Study-specific odds ratios (ORs) were correlated with the attack rates for each capsular serotype. A line of best fit was calculated using predictors from a linear regression based on the log-transformed ORs. The 2 variables are significantly correlated (P < .0001). Capsular serotypes 19A, 19F, 9A, 9N, 9V, 1, and 5 are excluded from the analysis, because for the first 5 of these capsular serotypes only serogroup data were available and there were large differences in the attack rates of the capsular serotypes within these serogroups. Capsular serotypes 1, 5, and 9A are crude estimates, because these capsular serotypes were isolated only from patients with invasive pneumococcal disease (IPD), but not from the nasopharynx. The duration of carriage for these capsular seroptypes was assumed to be <4 weeks, and the attack rate was estimated to be >60 cases/100,000. Those capsular serotypes that fall above the line represent capsular serotypes in which the invasive odds ratio has been overestimated in relation to the attack rates. Conversely, those capsular serotypes below the line are those that have been underestimated. Capsular serotypes in the heptavalent vaccine are indicated in bold type

Figure 3

Relationship between the invasive odds ratio derived from cross-sectional data and the attack rate measured directly from longitudinal data. Study-specific odds ratios (ORs) were correlated with the attack rates for each capsular serotype. A line of best fit was calculated using predictors from a linear regression based on the log-transformed ORs. The 2 variables are significantly correlated (P < .0001). Capsular serotypes 19A, 19F, 9A, 9N, 9V, 1, and 5 are excluded from the analysis, because for the first 5 of these capsular serotypes only serogroup data were available and there were large differences in the attack rates of the capsular serotypes within these serogroups. Capsular serotypes 1, 5, and 9A are crude estimates, because these capsular serotypes were isolated only from patients with invasive pneumococcal disease (IPD), but not from the nasopharynx. The duration of carriage for these capsular seroptypes was assumed to be <4 weeks, and the attack rate was estimated to be >60 cases/100,000. Those capsular serotypes that fall above the line represent capsular serotypes in which the invasive odds ratio has been overestimated in relation to the attack rates. Conversely, those capsular serotypes below the line are those that have been underestimated. Capsular serotypes in the heptavalent vaccine are indicated in bold type

Discussion

The main finding of the present study is that attack rate varies by capsular serotype, from >80 cases/100,000 acquisitions to <10 cases/100,000 acquisitions. Further, the duration of carriage for capsular serotypes for which there had been >10 acquisitions varied from 5.9 weeks for capsular serotype 15C to 19.9 weeks for capsular serotype 6B, and there was a significant inverse correlation between attack rate and duration of carriage. Lastly, there was a significant correlation between the published invasive odds ratio and attack rate, which is particularly reassuring. The main concern about previous studies [1, 2] that found variation in the invasive odds ratio was that variation in duration of carriage may have explained the variation in invasive odds ratio. The observation reported here indicates that the variation in duration of carriage between capsular serotypes is small compared with variation in their incidence of invasive disease. The value of this observation is that cross-sectional (carriage) data, which is much easier to collect than acquisition data, can be reliably used to assess the invasiveness of capsular serotypes. This will be useful in characterizing capsular serotypes that are emerging as a result of the reordering of the pneumococcal population after pneumococcal-conjugate vaccination

The main limitation of this study was that the sampling interval was at least 2 weeks. This limited detection of capsular serotypes with short durations of carriage (i.e., <2 weeks). This probably most affected the investigation of attack rate or invasive odds ratio for capsular serotypes 1, 5, and 9A. These capsular serotypes are clearly circulating within the children sampled for this study, because they contribute to disease in this population [6, 7]; however, acquisition was not detected with the sampling interval used. As such, only crude estimates of the attack rates for these capsular serotypes are presented. Furthermore, the sampling interval made it impossible to rule out the unlikely possibility that the long durations of carriage exhibited by some capsular serotypes were actually the result of capsular serotypes being lost and reacquired rather than being carried for long periods. However, if this did occur, it did not affect all capsular serotypes equally, which, therefore, strongly points to capsular serotype–specific difference

The observation here that duration of carriage varies by capsular serotype and that duration of carriage was inversely correlated with the attack rate is intriguing. This observation is consistent with capsular serotype determining not only how invasive a strain is but also how long it is carried. The explanatory mechanism for this capsular serotype–specific effect is not clear. It seems likely that it reflects a particular capsule-specific host interaction. Whether there is variation in duration of carriage within a capsular serotype that is dependent on genotype, as measured by multilocus sequence type (MLST), was not studied. No such variation was found by Brueggemann et al., who investigated the invasiveness of capsular serotypes [1]. Intriguingly, Sjöström et al. have recently shown that disease manifestation, morbidity, mortality, and risk of IPD, in adults, is associated with capsular serotype [10]. They were unable to demonstrate that genotype (MLST) played a role. Sjöström et al.’s data suggest that capsular serotypes that are carried for a short duration and have a high attack rate (or invasive odds ratio) are associated with milder disease and lower mortality and occur in people with little comorbidity. By contrast, capsular serotypes that are carried for a long duration and have a low attack rate are associated with more-severe disease and higher mortality and occur in people with more comorbidity. Consequently, capsular serotype may account for much of the host-parasite relationship in humans; but this interpretation must be tempered by the limited statistical power of some of these associations

Studies, mainly with mice, however, suggest that multiple genes other than capsule play a major role in pneumococcal virulence and invasiveness [11–14]. These animal studies would suggest that genotype may play a major role in determining the invasiveness of pneumococci; however, this has not been directly shown in humans. The extent to which the behavior of pneumococcal capsular serotypes in humans is determined by capsule versus other bacterial traits is unresolved, but knowing their relative contributions would assist in predicting the continued effectiveness of the capsule-specific pneumococcal-conjugate vaccines

If capsule is overwhelmingly the dominant factor determining the invasiveness and other properties of pneumococci, such as carriage duration and disease manifestation, the&amp;rank order of invasiveness for capsular serotypes is likely to remain unchanged. Consequently, the effectiveness of conjugate vaccine is more likely to persist, and any changes in disease pattern will probably involve capsular serotypes with a relatively high attack rate or invasive odds ratio, such as capsular serotypes 8, 12F, and 19A, which are currently not included in pneumococcal-conjugate vaccine formulations. In contrast, if factors other than capsule play a major role in determining invasiveness, the future is likely to be less predicable. Recombination between invasive lineages currently represented by capsular serotypes contained in the pneumococcal-conjugate vaccine and lineages representing relatively noninvasive capsular serotypes not contained in the vaccine could lead to novel variants. These variants may result in highly invasive strains, if their progeny contain the noncapsule invasive factors in association with any of the capsules not contained in pneumococcal-conjugate vaccine formulations. In either case, investigating the invasive odds ratios would provide insights into the behavior of these variants, compared with those capsular serotype lineages found in the prevaccine era, and information on whether, in humans, factors other than capsule play a major role in invasiveness. A major role for other factors would be likely if capsular serotypes that previously had a low invasive odds ratio exhibited substantially increased invasive odds ratios and occurred in association with novel genotypes. This possibility could be investigated, because a capsular serotype 19A variant with an MLST exclusively found among capsular serotype 4 strains has emerged in the United States since implementation of conjugate vaccine [15]. The question is whether the invasive odds ratio of this novel recombinant of 19A is equivalent to the 4-fold higher invasive odds ratio typical of capsular serotype 4 genetic lineages or whether it has fallen to that of capsular serotype 19A lineages

Acknowledgments

We thank the parents and infants who gave their time and participated in the studies. We also thank the staff of the Oxford Vaccine Group and all the microbiologists involved in running the active surveillance. We also thank Angela Brueggemann for providing invasive odds ratio estimates. Robert George and Elizabeth Miller of the Health Protection Agency, Colindale, London, provided the national England and Wales data, for which we are grateful

References

1
Brueggemann
AB
Griffiths
DT
Meats
E
Peto
T
Crook
DW
Clonal relationships between invasive and carriage Streptococcus pneumoniae and serotype- and clone-specific differences in invasive disease potential
J Infect Dis
 , 
2003
, vol. 
187
 (pg. 
1424
-
32
)
2
Brueggemann
AB
Peto
TE
Crook
DW
Butler
JC
Kristinsson
KG
Temporal and geographic stability of the serogroup-specific invasive disease potential of Streptococcus pneumoniae in children
J Infect Dis
 , 
2004
, vol. 
190
 (pg. 
1203
-
11
)
3
Gray
BM
Converse GM 3rd
Dillon
HC
Jr
Epidemiologic studies of Streptococcus pneumoniae in infants: acquisition, carriage, and infection during the first 24 months of life
J Infect Dis
 , 
1980
, vol. 
142
 (pg. 
923
-
33
)
4
Gray
BM
Turner
ME
Dillon
HC
Jr
Epidemiologic studies of Streptococcus pneumoniae in infants: the effects of season and age on pneumococcal acquisition and carriage in the first 24 months of life
Am J Epidemiol
 , 
1982
, vol. 
116
 (pg. 
692
-
703
)
5
Sleeman
KL
Daniels
L
Gardiner
M
, et al.  . 
Acquisition of Streptococcus pneumoniae and nonspecific morbidity in infants and their families: a cohort study
Pediatr Infect Dis J
 , 
2005
, vol. 
24
 (pg. 
121
-
7
)
6
Miller
E
Waight
P
Efstratiou
A
Brisson
M
Johnson
A
Epidemiology of invasive and other pneumococcal disease in children in England and Wales 1996–1998
Acta Paediatr Suppl
 , 
2000
, vol. 
89
 (pg. 
11
-
6
)
7
Sleeman
K
Knox
K
George
R
, et al.  . 
Invasive pneumococcal disease in England and Wales: vaccination implications
J Infect Dis
 , 
2001
, vol. 
183
 (pg. 
239
-
46
)
8
O’Brien
KL
Nohynek
H
Report from a WHO Working Group: standard method for detecting upper respiratory carriage of Streptococcus pneumoniae
Pediatr Infect Dis J
 , 
2003
, vol. 
22
 (pg. 
e1
-
11
)
9
Lund
BC
Henrichsen
J
Bergen
D
Norris
J
Laboratory diagnosis, serology and epidemiology of Streptococcus pneumoniae
Methods in microbiology
 , 
1978
London
Academic Press
(pg. 
241
-
62
)
10
Sjöström
K
Spindler
C
Ortqvist
A
, et al.  . 
Clonal and capsular type decides whether pneumcococci will act as a primary or opportunistic pathogen
Clin Infect Dis
 , 
2006
, vol. 
42
 (pg. 
451
-
9
)
11
Brown
JS
Gilliland
SM
Holden
DW
A Streptococcus pneumoniae pathogenicity island encoding an ABC transporter involved in iron uptake and virulence
Mol Microbiol
 , 
2001
, vol. 
40
 (pg. 
572
-
85
)
12
Hava
DL
Camilli
A
Large-scale identification of serotype 4 Streptococcus pneumoniae virulence factors
Mol Microbiol
 , 
2002
, vol. 
45
 (pg. 
1389
-
406
)
13
Lau
GW
Haataja
S
Lonetto
M
, et al.  . 
A functional genomic analysis of type 3 Streptococcus pneumoniae virulence
Mol Microbiol
 , 
2001
, vol. 
40
 (pg. 
555
-
71
)
14
Sandgren
A
Albiger
B
Orihuela
CJ
Tuomanen
E
Normark
S
Virulence in mice of pneumococcal clonal types with known invasive disease potential in humans
J Infect Dis
 , 
2005
, vol. 
192
 (pg. 
791
-
800
)
15
Pai
R
Moore
MR
Pilishvili
T
Gertz
RE
Whitney
CG
Postvaccine genetic structure of Streptococcus pneumoniae serotype 19A from children in the United States
J Infect Dis
 , 
2005
, vol. 
192
 (pg. 
1988
-
95
)
Potential conflicts of interest: D.W.C. is a recipient of a grant from Wyeth Pharmaceuticals and Aventis Pharmaceuticals
Financial support: Wellcome Trust (grant 056886/2/994/Z); D.W.C. was a Wellcome Trust Leave Fellow while undertaking this study (grant 057366/Z/99/Z)