Serum Serotype-Specific Pneumococcal Anticapsular Immunoglobulin G Concentrations after Immunization with a 9-Valent Conjugate Pneumococcal Vaccine Correlate with Nasopharyngeal Acquisition of Pneumococcus

Abstract

BackgroundImmunization with pneumococcal conjugate vaccines (PCVs) reduces nasopharyngeal colonization by Streptococcus pneumoniae. We attempted to correlate postvaccination serum serotype-specific pneumococcal anticapsular immunoglobulin (Ig) G concentrations with new acquisitions of vaccine-type (VT) serotypes and the VT-related serotype 6A

MethodsA total of 132 day care center attendees aged 12–35 months received a 9-valent PCV (PnCRM9) and were followed for 2 years for new nasopharyngeal acquisitions of S. pneumoniae. A total of 132 control subjects received a meningococcus type C conjugate vaccine. Serum serotype-specific pneumococcal anticapsular IgG concentrations were determined at 1 month after complete immunization

ResultsA logistic regression model of the probability of having a new acquisition of S. pneumoniae (for serotypes 9V, 14, 19F, and 23F) as a function of the IgG concentration showed a negative coefficient, indicating that higher IgG concentrations led to a decreasing probability of having a new acquisition, and achieved statistical significance for serotypes 14 and 19F. Similarly, a new acquisition of serotype 6A was shown to be significantly inversely related to the anti-6B IgG concentration. An effect of the IgG concentration on duration of carriage was not demonstrated

ConclusionThe magnitude of herd protection against S. pneumoniae provided by a PCV may depend on the magnitude of IgG concentrations

Immunization with pneumococcal conjugate vaccines (PCVs) reduces nasopharyngeal carriage of Streptococcus pneumoniae serotypes included in the vaccine (vaccine-type [VT] serotypes) and serotypes immunogenically related to those in the vaccine in infants and toddlers [1–10 ]. PCVs reduce the carriage rate by preventing new acquisitions of S. pneumoniae and not by shortening the duration of existing colonization [9, 11]. This phenomenon, although well established, is not completely understood, because the attachment of S. pneumoniae to mucosal surfaces of mammalian tissues is not mediated by any known mechanism that involves the polysaccharide capsule of these organisms [12, 13]. One frequently cited speculation for the mechanism of this phenomenon is that, when present in sufficient amounts, antibodies bind to the polysaccharide capsule and create a steric inhibition of the interaction of pneumococcal surface proteins with binding sites on the mammalian epithelial cell surface

Previous studies have suggested that colonization is prevented, not by secretory pneumococcal anticapsular IgA, but by the presence, in saliva, of pneumococcal anticapsular IgG that either leaks from the serum or is secreted by committed plasma cells present on mucosal surfaces [14]. Furthermore, pneumococcal anticapsular IgG concentrations in saliva are related to these concentrations in serum, and this supports the hypothesis that salivary pneumococcal anticapsular IgG derives from the serum [15–17 ]. Secretory IgA plays only a minor role in protection from colonization, possibly because S. pneumoniae has specific proteins that cleave secretory IgA but not IgG [18, 19]

If this speculation accurately characterizes the mechanism involved, then the inhibition of colonization would depend on serum IgG concentrations. Therefore, we attempted to correlate the postvaccination serum serotype-specific pneumococcal anticapsular IgG concentrations with new acquisitions of VT and VT-related S. pneumoniae serotypes. For this purpose, we investigated children attending day care centers (DCCs) who were vaccinated with a 9-valent PCV conjugated to diphtheria toxin CRM₁₉₇ (PnCRM9)

Subjects, Materials, and Methods

SubjectsHealthy boys and girls aged 12–35 months attending 8 DCCs in Beer-Sheva in southern Israel were recruited. The subjects were enrolled in a randomized study on the effect that the PnCRM9 vaccine has on nasopharyngeal carriage of S. pneumoniae in toddlers attending DCCs. The detailed methods and some of the results have been published elsewhere [9, 20, 21]. In the present study, we attempted to determine the role that serum serotype-specific pneumococcal anticapsular IgG concentrations play in the prevention of new acquisitions of S. pneumoniae serotypes in vaccinees. The subjects who received the control vaccine (a meningococcus type C conjugate vaccine coupled to CRM₁₉₇) were used in this study as a reference group as a whole, and there was no further analysis of the effect that serum serotype-specific pneumococcal anticapsular IgG concentrations have on new acquisitions of S. pneumoniae serotypes in this group

Subjects were excluded if they (1) had received or were expected to receive any vaccine or immunoglobulin during the 4-week period before or during the 4-week period after the administration of the study vaccines, (2) had any known or suspected impairment of immunologic functions, (3) had a major congenital malformation or serious chronic disease, (4) had a known hypersensitivity to any component of the study vaccines or had any previous severe vaccine-associated adverse reaction, or (5) had been previously vaccinated with any pneumococcal or meningococcal vaccine. Subjects were also temporarily excluded if they had had any febrile illness (rectal temperature, ⩾38°C) within 72 h before vaccination

VaccinesThe vaccine being studied was PnCRM9, a 9-valent PCV in which 2 μg each of S. pneumoniae serotypes 1, 4, 5, 9V, 14, 18C, 19F, and 23F carbohydrates and 4 μg of serotype 6B carbohydrate were coupled to diphtheria toxin CRM₁₉₇ and were presented as a lyophilized preparation (Wyeth-Lederle Vaccines; WLV). The control vaccine was a meningococcus type C conjugate vaccine in which 10 μg of carbohydrate was coupled to CRM₁₉₇

Study designA total of 264 subjects were randomized to receive either the PnCRM9 vaccine (n=132) or the control vaccine (n=132). Subjects aged 12–17 months at the time of enrollment (n=19 subjects who received the PnCRM9 vaccine; n=16 subjects who received the control vaccine) received 2 intramuscular injections 2–3 months apart, and those aged 18–35 months at the time of enrollment (n=113 subjects who received the PnCRM9 vaccine; n=116 subjects who received the control vaccine) received a single intramuscular injection. For each subject, study visits were planned that were to occur monthly during the first year and every 2 months during the second year of follow-up. At each visit, a nasopharyngeal sample was obtained for culture for S. pneumoniae. Blood samples were obtained beginning at 1 month after complete immunization (1 month after the second dose for those aged 12–17 months at the time of enrollment and 1 month after the single dose for those aged 18–35 months at the time of enrollment). The code of this double-blind study was broken at the end of follow-up, after the last sample was analyzed in the laboratory

Serological testingAll blood samples were refrigerated immediately after the blood was drawn. Serum was separated within 16 h, and the samples were frozen at −70°C until further processing. Serum IgG responses to serotypes 4, 6A, 6B, 9V, 14, 18C, 19F, and 23F were determined by ELISA, as described elsewhere [22]

Nasopharyngeal culturesNasopharyngeal samples were obtained by use of a flexible dacron-tipped swab, which was introduced into the nostrils and was advanced until resistance was found. These swabs were inoculated into modified Stewart transport medium (Medical Wire & Equipment) and processed within 1 h at the Clinical Microbiology Laboratory of the Soroka University Medical Center (Beer-Sheva, Israel). Material from swabs was plated on Columbia agar with 5% sheep’s blood and 5.0 μg/mL gentamicin, and the plates were incubated aerobically at 35°C in a CO₂-enriched atmosphere for 48 h. This method was used in our previous studies and yielded a high rate of positive cultures [1, 2, 23–25 ]

The presumptive identification of S. pneumoniae was based on the presence of α-hemolysis and the inhibition by optochin, and the identity of the bacteria present was confirmed by a positive slide agglutination test result (Phadebact; Pharmacia Diagnostics). One S. pneumoniae colony per plate was then subcultured, harvested, and kept frozen at −70°C for further testing

Serogrouping and serotyping.Serogrouping and serotyping of S. pneumoniae were performed by the quellung reaction, using serum samples produced by the Statens Seruminstitut [26]. Isolates that had negative reactions to all pooled serum samples and to omni serum were considered to be nontypeable. Serotypes 1, 4, 5, 6B, 9V, 14, 18C, 19F, and 23F were classified as VT S. pneumoniae serotypes

Data management and statistical analysisThe primary end point for the study was a new acquisition of any VT S. pneumoniae serotype. A new acquisition was defined as the detection of a serotype, once a subject was fully vaccinated, that had not been detected previously, including in the prevaccination nasopharyngeal sample. Therefore, only 1 new acquisition could be counted per serotype for each subject, and samples only from subjects who received the PnCRM9 vaccine were analyzed in the logistic regression analysis

Analysis of relationship between serum serotype-specific pneumococcal anticapsular IgG concentrations and new acquisitions of S. pneumoniaeserotypes The data analyzed were the first occurrence of carriage during the 2 years of follow-up. The probability of having a new acquisition was modeled using logistic regression analysis. Carriage of serotype 6A was modeled using serologic values for serotype 6B

Analysis of duration of carriage in relation to serum serotype-specific pneumococcal anticapsular IgG concentrations For each subject, an occurrence of carriage for a serotype was counted if it occurred after the final vaccine dose. The duration of carriage was calculated by adding, for each subject, the number of visits when carriage of a specific serotype was detected. This method of determining the duration of carriage does not differentiate between consecutive episodes and separate episodes. For example, a serotype detected at 2 visits that were 6 months apart would be calculated as having a duration of carriage of 2 months, as would a serotype that was detected during 2 consecutive months. Because the interval between visits varied between the first (every month) and the second (every 2–3 months) year of follow-up, data from the first and second years of follow-up were analyzed separately

Results

A total of 132 subjects who received the PnCRM9 vaccine and 132 subjects who received the control vaccine were enrolled (table 1). Of the expected 2147 study visits by the subjects who received the PnCRM9 vaccine, 1840 (86%) study visits occurred; of the expected 2106 study visits by the subjects who received the control vaccine, 1813 (86%) visits occurred. Serum samples obtained 1 month after complete immunization were available for IgG testing from 129 subjects who received the PnCRM9 vaccine and from 126 subjects who received the control vaccine

Table 1

Demographic characteristics and nasopharyngeal samples obtained during follow-up for 129 subjects who received the PnCRM9 vaccine and 126 subjects who received the control vaccine

Relationship between serum serotype-specific pneumococcal anticapsular IgG concentrations and new acquisitions of S. pneumoniaeserotypes   Of the 129 subjects who received the PnCRM9 vaccine, 19 (15%) were aged 12–17 months at the time of enrollment (and therefore received 2 doses of the vaccine). For these 129 subjects, the durations of follow-up were <6, 6–11, 12–17, and 18–29 months for 4, 7, 4, and 114 subjects, respectively. Because only 5 colonization events of interest were found in the 15 subjects with shorter durations of follow-up, all subjects were included in the analysis, regardless of the durations of their follow-up. The effect of these shorter durations of follow-up was examined by removing from the data for these 15 subjects another subject’s colonizing serotype, even if no colonization was found. No change in conclusions resulted from this exploratory analysis

Of the 129 subjects who received the PnCRM9 vaccine, 102 (79%) carried S. pneumoniae at the time of enrollment; of these 102 subjects, 48 (47%) carried 1 VT serotype, and 22 (22%) carried serotype 6A. Of the 126 subjects who received the control vaccine, 101 (80%) carried S. pneumoniae at the time of enrollment; of these 101 subjects, 51 (50%) carried 1 VT serotype, and 18 (18%) carried serotype 6A. In the subjects who received the PnCRM9 vaccine, during the first and second years of follow-up, 69 and 32 new acquisitions of the 9 VT S. pneumoniae serotypes occurred, respectively (table 2). In the subjects who received the control vaccine, 121 and 62 new acquisitions of the 9 VT S. pneumoniae serotypes occurred, respectively. Because a limited number of new acquisitions of VT S. pneumoniae serotypes in the subjects who received the PnCRM9 vaccine occurred, new acquisitions of serotypes 9V, 14, 19F, and 23F only were analyzed

Table 2

Number of new acquisitions of vaccine-type (VT) Streptococcus pneumoniae serotypes and serotype 6A during follow-up in 129 subjects who received the PnCRM9 vaccine and 126 subjects who received the control vaccine, by serotype

At 1 month after vaccination, serum serotype-specific pneumococcal anticapsular IgG concentrations for each of the analyzed serotypes showed a wide variability in both the subjects who received the PnCRM9 vaccine and the subjects who received the control vaccine (table 3 and figure 1). In general, in the subjects who received the PnCRM9 vaccine, the serum serotype-specific pneumococcal anticapsular IgG concentrations in those who were aged 12–17 months at the time of vaccination (and therefore received 2 doses of vaccine) were higher than those of other age groups (who received only 1 dose of vaccine), both at 1 month after vaccination and at 1 year after vaccination

Table 3

Serum serotype-specific pneumococcal anticapsular IgG concentrations for Streptococcus pneumoniae serotypes 6B, 9V, 14, 18C, 19F, and 23F in 129 subjects who received the PnCRM9 vaccine and in 126 subjects who received the control vaccine at 1 month after complete immunization, by age at the time of immunization

Figure 1

Reverse cumulative distribution curves of serum serotype-specific pneumococcal anticapsular IgG, for Streptococcus pneumoniae serotypes 6B, 9V, 14, 18C, 19F, and 23F, 1 month after complete immunization in subjects who received the PnCRM9 vaccine (VAC; n=129) and subjects who received the control vaccine (CON; n=126)

The logistic regression model of the probability of having a new acquisition of each S. pneumoniae serotype (9V, 14, 19F, and 23) as a function of the serum serotype-specific pneumococcal anticapsular IgG concentration was developed as follows. First, a logistic regression model was fit with the terms age group; serotype; the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations; and the interactions between age and serotype, age and the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations, and serotype and logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations. There were no significant (P>.05) effects of the age variables. The model was then fit without the age variables, and the following P values were found for effect: serotype, P<.0001; the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations, P=.0241; and the interaction between serotype and the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations, P=.397. Therefore, the association of the odds of having a new acquisition and the serum serotype-specific pneumococcal anticapsular IgG concentration was not significantly different between serotypes. Then, the model was refit without the interaction term, and the following results were found: for the effect of serotype, P<.001; estimate of the coefficient of the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentration, −0.58 (95% confidence interval [CI], −0.95 to −0.22). Therefore, increasing serum serotype-specific pneumococcal anticapsular IgG concentrations led to a decreasing probability of having carriage, and, when adjusted for serum serotype-specific pneumococcal anticapsular IgG concentrations, the probability of having carriage was different between serotypes. The model was also refit individually for each serotype (table 4). Although all serotypes had negative coefficients, which indicates that the probability of having a new acquisition decreases with increasing serum serotype-specific pneumococcal anticapsular IgG concentrations, statistical significance was reached only for serotypes 14 and 19F. The actual percentages of new acquisitions (in 1-log increments) and the percentages of new acquisitions predicted by the individual serotype logistic regression model are depicted in figure 2

Table 4

Results of the logistic regression model of the probability of having a new acquisition of Streptococcus pneumoniae serotypes 9V, 14, 19F, and 23F during follow-up in 129 subjects who received the PnCRM9 vaccine

Figure 2

Percentages of new acquisitions (in 1-log increments) and percentages of new acquisitions predicted by the logistic regression model, by Streptococcus pneumoniae serotype (9V, 14, 19F, and 23F), in subjects who received the PnCRM9 vaccine (black squares) and in subjects who received the control vaccine (black circle)

In our study, the most commonly newly acquired S. pneumoniae serotype was 6A (table 2). This serotype, which was not included in the PnCRM9 vaccine, cross-reacts with serotype 6B, which was included in the PnCRM9 vaccine. We applied the logistic regression model for the probability of having a new acquisition of serotype 6A as a function of the serum anti-6B IgG concentration. When this model was applied, the coefficient of the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentration was −0.63 (P = .009; 95% CI, −1.14 to −0.15). Figure 3 depicts the actual percentages of new acquisitions of serotype 6A (in 1-log increments of serum anti-6B IgG concentration) and the percentages of new acquisitions predicted by the logistic regression model. Thus, an effect of the serum serotype-specific pneumococcal anticapsular IgG concentration elicited by the administration of serotype 6B on a new acquisition of serotype 6A was demonstrated

Figure 3

Percentages of new acquisitions (in 1-log increments) and percentages of new acquisitions predicted by the logistic regression model of Streptococcus pneumoniae serotype 6A (using serotype 6B serologic values) in subjects who received the PnCRM9 vaccine (black squares) and in subjects who received the control vaccine (black circle)

Relationship between serum serotype-specific pneumococcal anticapsular IgG concentrations and duration of carriage An analysis of the relationship between serum serotype-specific pneumococcal anticapsular IgG concentrations and duration of carriage was performed for serotypes 6B, 9V, 14, 18C, 19F, and 23F only (because of the small number of new acquisitions of serotypes 1, 4, and 5). The linear regression analysis of duration of carriage was performed by year. In each of the 2 regression analyses, a model was fit with the main effects of serotype and the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations 1 month after vaccination. Although the coefficient of the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations was slightly negative in each of the regression analyses (−0.21 and −0.16 for the first and second years of follow-up, respectively), which indicates that there was a shorter duration of carriage with increasing serum IgG concentrations, for none of the regression analyses was P<.30 for the serum IgG concentration. Thus, an effect of serum IgG concentration on duration of carriage could not be demonstrated

Discussion

In the present study, we demonstrated that, after vaccination with PCVs, the number of new acquisitions of S. pneumoniae serotypes 14 and 19F was inversely correlated with the serum serotype-specific pneumococcal anticapsular IgG concentration. Because of the relatively small sample size, we were not able to demonstrate a statistically significant effect for other VT serotypes individually, and, because of the scarcity of the observed new acquisitions, we could not even attempt to calculate the effect of the vaccine for some others. However, it is likely that, biologically, these serotypes will act like those for which the number of new acquisitions was sufficient for our analysis

This effect, although demonstrated after vaccination with a PCV, may be only 1 of several mechanisms that prevent new acquisitions of S. pneumoniae serotypes in maturing unvaccinated individuals, because antibodies to various surface proteins, acquired after pneumococcal colonization or infection, may also be protective [13, 27, 28]. In addition, the role that nasopharyngeal flora play at various ages and the potential modification of the human epithelial cell receptors with age may also affect the risk of colonization. For this reason, we chose not to analyze having a new acquisition as a function of serum serotype-specific pneumococcal anticapsular IgG concentrations in the subjects who received the control vaccine but used them as a reference group only

When the effect of the PnCRM9 vaccine on duration of carriage once a serotype was acquired was analyzed, the coefficient of the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations was negative in each of the regression analyses, which suggests that there is a shorter duration of carriage with increasing serum IgG concentrations. However, the coefficient of the logarithm of the serum serotype-specific pneumococcal anticapsular IgG concentrations was only slightly negative in all regression analyses, and no statistical significance was shown in any of the studied serotypes, which suggests that, if there is any effect of the PCV on duration of carriage, it is a marginal one. This observation is similar to that made elsewhere with regard to Haemophilus influenzae type b (Hib) conjugate vaccines [11]. The reason for this finding is not clear, but it can be speculated that high serum serotype-specific pneumococcal anticapsular IgG concentrations prevent colonization by creating a steric inhibition of the interaction of pneumococcal surface proteins with binding sites, but once carriage is established and the binding proteins are in contact with the host’s cells, the potential steric inhibition by serum pneumococcal anticapsular IgG is no longer effective

Another important observation in our study was the suggested ability of IgG elicited by 1 serotype to cross-protect from colonization by another serotype in the same group, as was demonstrated by the reduction in the number of new acquisitions of serotype 6A with an increase in the serum anti-6B IgG concentration. Although cross-protection against serotype 6A by immunization with serotype 6B has been previously demonstrated in invasive infections [29, 30], mucosal infections such as acute otitis media (AOM) [31, 32], and even carriage [9], this is the first study, to our knowledge, that demonstrates in humans the correlation between the serum IgG concentration against 1 serotype in a group and the prevention of a new acquisition of a cross-reacting serotype

The findings in this study have important implications. A long-standing debate exists with regard to the relative importance of the serum serotype-specific pneumococcal anticapsular IgG concentration versus the priming effect of previous doses in protection against invasive infections after the administration of conjugate vaccines. It is plausible that the most important mechanism in the prevention of invasive Hib or pneumococcal infections is priming and that excellent protection against invasive infection can be achieved even in the presence of low concentrations of circulating antibodies [33]. Recent data from the United Kingdom partly challenge this hypothesis. An increase in invasive Hib disease incidence was observed when a Hib conjugate vaccine was administered to infants at short intervals in combination with acellular pertussis vaccine, without booster administration, and this led to very low concentrations of circulating anti–polyribosylribitol phosphate (PRP) antibodies [34, 35]. This increase of disease was observed despite the ability of the vaccinees to respond rapidly with high serum anti-PRP IgG concentrations to a booster dose

With regard to mucosal infections such as AOM, studies have demonstrated that protection from disease may be correlated not only with priming but also with specific humoral pneumococcal anticapsular IgG concentrations. In this regard, data from 2 efficacy studies conducted in Finland on vaccination with either a 7-valent PCV conjugated to CRM₁₉₇ (PnCRM7) [31] or a 7-valent PCV conjugated to an outer membrane protein complex of Neisseria meningitidis B (PnOMPC7) in patients with AOM [36] are in accordance with our data on nasopharyngeal carriage. First, when results of the 2 Finnish studies were combined, a clear correlation between IgG concentrations 4 weeks after vaccination and protection against AOM caused by serotypes 6B, 19F, and 23F was found [37]

Second, protection against serotype 19F was achieved only with serum IgG concentrations that were much higher than those with which protection was achieved against the other serotypes. It should be noted that these correlations are with serum IgG at its peak concentrations after immunization, not with serum IgG concentrations at the time of a new acquisition, which are likely to be much lower. Our findings, although not conclusive for all serotypes (probably because of small sample size) are consistent with serum serotype-specific pneumococcal anticapsular IgG concentration–dependent protection (figure 2). In this regard, the risk for having a new acquisition of serotype 19F seemed to be higher for each serum IgG concentration range than that with the other serotypes. Also, we cannot rule out the possibility that the association between serum IgG concentration and the risk of having a new acquisition might not exist for some serotypes

Third, although both vaccines used in the Finnish study (PnCRM7 and PnOMPC7) resulted in high efficacy against AOM caused by the VT serotypes, the PnCRM7 vaccine had an efficacy of 51% (95% CI, 44%–66%) against the cross-reacting serotypes, whereas the PnOMPC7 vaccine had an efficacy of −5% (95% CI, −47% to 25%) for these serotypes. The lower serum IgG concentrations in the PnOMPC7 group could explain the vaccine’s failure to protect against serotypes that cross-react with the VT serotypes [36]. This hypothesis is supported by the results of a study by Vakevainen et al. [38], who demonstrated the need for 2–6 times more anti-6B antibodies to support 50% opsonophagolytic killing of serotype 6A than was needed to support the same level of killing of serotype 6B. When the results of the 2 efficacy studies were combined, serum anti-6B IgG concentrations correlated with protection against 6A but showed that higher serum IgG concentrations were needed than those needed to protect against serotype 6B. This finding is also in accordance with that in our study, in which prevention of colonization by serotype 6A correlated with serum anti-6B IgG concentrations

Thus, a vaccine that elicits higher serum IgG responses may be more protective against AOM or other mucosal infections caused by VT serotypes than a vaccine that elicits lower serum IgG responses. Our study shows that achieving higher serum pneumococcal anticapsular IgG concentrations has a substantial effect not only on the prevention of disease, as was shown by Jokinen et al. [37], but also on the prevention of carriage of S. pneumoniae in healthy individuals. This was previously suggested by Sigurdardottir et al. [39]. They immunized children with an 11-valent PCV consisting of a mixture of diphtheria and tetanus toxoid conjugates. They showed that nasopharyngeal carriage of VT S. pneumoniae serotypes was more common in children with poor responses to the vaccine than in children with stronger responses

Our study has 2 limitations. First, the reference serum IgG concentration was obtained from a single measurement. It is likely that serum IgG concentrations either decreased with time or increased after contact with carriers. This was not taken into account. Second, in regard to the detection of new acquisitions, we were limited by the intervals between study visits (1–3 months) and the sensitivity of the cultures. Thus, we suspect that new acquisitions occurred more frequently than they were detected in our study. However, both limitations existed for both arms of the study. Furthermore, these limitations exist for all studies of this kind

Our findings suggest that, in comparing 2 PCVs, the vaccine that elicits higher serum serotype-specific pneumococcal anticapsular IgG responses may have a stronger effect on pneumococcal carriage, because reduction of carriage in vaccinated children is achieved mainly by the prevention of new acquisitions [9]. Thus, these findings have important implications in the United States, where the widespread administration of the PnCRM7 vaccine resulted in the prevention of invasive pneumococcal disease not only in vaccinees but also in unvaccinated control subjects, especially those aged 2–4 years (representing the age group of the older siblings of our study participants), aged 20–39 years (representing the age group of the parents), and aged ⩾65 years (representing the age group of the grandparents). This phenomenon, termed “herd immunity,” can be explained only by reduction of carriage and by a consequent reduction of transmission of S. pneumoniae from the vaccinated children to their contacts [20]. Thus, the magnitude of herd immunity against S. pneumoniae achieved by vaccination with PCVs may depend on the magnitude of serum IgG concentrations achieved by using specific PCVs and the administration of a booster dose

Presented in part: 41st Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, 16–19 December 2001 (abstract G-2040)

Financial support: Wyeth-Lederle Vaccines

Potential conflicts of interest: R.D. has received research funding and is on the advisory board of Wyeth-Lederle Vaccines; G.R.S. is an employee of Wyeth-Lederle Vaccines; R.K. is a consultant for Wyeth Vaccine Research

Present affiliation: Blair & Company, Greenwich, Connecticut