Long-Term Effect of Pneumococcal Conjugate Vaccine on Nasopharyngeal Colonization by Streptococcus pneumoniae—and Associated Interactions with Staphylococcus aureus and Haemophilus influenzae Colonization—in HIV-Infected and HIV-Uninfected Children

Abstract

After a primary series of 3 doses, it was found that a 9-valent pneumococcal conjugate vaccine no longer reduces nasopharyngeal colonization by vaccine serotypes in childern 5.3 years of age. In addition, human immunodeficiency virus (HIV-infected children (n= 81) had a higher prevalence of colonization by Streptococcus pneumoniae and Haemophilus influenzae (71.6% and 74.1%, respectively) than did HIV-uninfected children (n = 271; 50.9% and 52.0%, respectively), suggesting that increased colonization may contribute to the greater burden of pneumococcal disease in HIV-infected children. Inverse associations between colonization by S. pneumoniae and colonization by Staphylococcus aureus and between colonization by S. aureus and colonization by H. influenzae were observed only in HIV-uninfected children, possibly as a result of suboptimal adaptive immunity after previous colonization in HIV-infected children.

Both 7-valent and 9-valent serotype-specific pneumococcal polysaccharide-CRM₁₉₇ conjugate vaccines (PCV-7 and PCV-9, respectively) reduce nasopharyngeal colonization by vaccine serotypes, and increase colonization by nonvaccine serotypes, within months following receipt of the primary vaccine series [1]. There are contradictory reports on the durability of protection against nasopharyngeal acquisition of vaccine-serotype pneumococci [2, 3]. To our knowledge there are no published data on long-term pneumococcal colonization in the absence of a booster dose of any pneumococcal vaccine. In addition, nasopharyngeal colonization by vaccine-serotype pneumococci has been reported to be inversely associated with colonization by Staphylococcus aureus in human immunodeficiency virus (HIV)—uninfected children [4, 5].

This study aimed to (1) describe the long-term effect of PCV-9 on vaccine-serotype—specific colonization in HIV-uninfected and HIV-infected children in the absence of a booster dose of PCV, (2) compare the prevalence of pneumococcal colonization between HIV-uninfected and HIV-infected children, and (3) determine the interactions of nasopharyngeal colonization by Streptococcus pneumoniae, S. aureus, and Haemophilus influenzae in HIV-uninfected and HIV-infected children.

Subjects, materials, and methods. Between March 2001 and April 2005, a total of 3797 of 8000 randomly invited parents of children who had participated in a phase 3 efficacy trial between March 1998 and November 2001 [6] responded to an invitation to have their children receive a dose of PCV-7. The children had been previously individually randomized, at 6, 10, and 14 weeks of age, to receive either 3 doses of PCV-9 (containing 2 μg each of serotypes 1, 4, 5, 9V, 14, 19F, and 23F polysaccharide; 2 μg of serotype 18C oligosaccharide; and 4 μg of serotype 6B polysaccharide, each conjugated individually to the protein CRM₁₉₇) or a placebo. A random subset of children who had received 3 doses of study vaccine per protocol during the trial were identified, and their parents were requested to consider their child's participation in this study. Study personnel remained blinded to the initial randomization of the children.

Nasopharyngeal swabs were collected using a Dacron-tipped swab on a flexible aluminium shaft (Cat 151D; Medical Wire Equipment), inoculated into skim milk tryptone-glucoseglycerin transport medium and processed at the Respiratory and Meningeal Pathogens Research Unit laboratory in South Africa. Serotyping of identified pneumococci was performed using the quellung method (Statens Serum Institute).

S. aureus and H. influenzae were isolated by plating out 100 μL of the skim milk tryptone-glucose-glycerin medium onto man1662 nitol salt agar (Mast Group) and onto chocolate agar, respectively, supplemented with 150 μg/mL of bacitracin (Sigma Aldrich). Mannitol-fermenting staphylococci were identified as S. aureus on the basis of the positive test results for catalase and coagulase. Isolates of H. influenzae were identified on the basis of their nutritional dependence on both hemin and nicotine adenine nucleotide. No booster dose of any pneumococcal vaccine was given to the children before nasopharyngeal swabs were obtained.

Data were analyzed using STATA version 8.0 and Epi Info version 6.04d. Continuous data were compared using Student's t test. Proportions were compared using either a χ² test or a 2-tailed Fisher's exact test, as appropriate. P ⩽ .05 was considered to be statistically significant. Relative risks (RRs) and 95% confidence intervals (CIs) were used to compare differences in nasopharyngeal colonization. HIV-infected children were clinically and immunologically categorized using the 1994 Centers for Disease Control and Prevention (CDC) recommendations. A Cochran-Armitage χ² test was performed to assess the trend of colonization prevalence in HIV-infected children during different clinical stages (asymptomatic [N]; mildly symptomatic [A], moderate AIDS [B], and severe AIDS [C]) and over strata of HIV loads (<10,000, 10,000–100,000, and >100,000 copies/mL). Because of the limited number of children, a comparison of colonizations in relation to the CD4⁺ immunological stage was restricted to children with CD4₊ T-lymphocyte percentages of <15% versus those with ⩾15%. Any serogroup represented in PCV-9 was considered to be a vaccine serogroup. The Ethics Committee for Research on Humans Subjects at the University of the Witwatersrand approved the study, and the parents of the children gave written consent.

Results. A total of 352 children (153 PCV-9 recipients and 199 placebo recipients; 271 HIV uninfected, 81 HIV infected) were enrolled. The mean age of the children was 5.64 years (SD, 0.79 years), and nasopharyngeal samples were collected 5.35 years (SD, 0.81 years) after the third dose of the study vaccine was administered; neither the collection time of the nasopharyngeal samples nor the third dose of the study vaccine differed between groups (table 1). The clinical stage, the immunological stage, and the viral-load category did not differ between HIV-infected PCV-9 and placebo recipients. The clinical CDC categorization of the HIV-infected children included 9 (11.1%) at stage N, 13 (16.0%) at stage A, 51 (63.0%) at stage B, and 8 (9.9%) at stage C. Of the 32 HIV-infected children for whom CD4⁺percentage counts were available, 26 were <15%, 4 were 15%–25%, and 2 were >25%. The HIV load was available for 43 children and was <10,000 copies/mL in 17 children, 10,000–100,000 copies/mL in 13 children, and >100,000 copies/mL in the remaining 13 children. Fifteen (18.5%) of the HIV-infected children were receiving highly active antiretroviral therapy (ART).

Table 1.

Demographic information on recipients of 9-valent pneumococcal conjugate vaccine (PCV-9) and recipients of placebo studied to determine the long-term effect that PCV-9 has on the prevalence of pneumococcal nasopharyngeal colonization in HIV-infected and HIV-uninfected children.

During the prevalence of vaccine-serotype or non—vaccine-serotype colonization, there was no difference between PCV-9 recipients and placebo recipients, regardless of whether they were stratified by HIV-infection status (table 2). HIV-infected children were 1.41-fold (95% CI, 1.17–1.68) more likely to be colonized by pneumococci than were HIV-uninfected children (71.6% vs. 50.9%, respectively; table 2), and this was particularly evident when vaccine serotypes were used (44.4% vs. 16.6%, respectively; mean RR, 2.68 [95% CI, 1.87–3.84]). The serotype distribution of pneumococcal isolates is available online (table 3).

Table 2.

Prevalence of colonization by Streptococcus pneumoniae colonization, Staphylococcus aureus colonization, and Haemophilus influenzae, 5.3 years after receipt of a primary series of 3 doses of 9-valent pneumococcal conjugate vaccine (PCV-9) or placebo, in HIV-infected and HIV-noninfected children.

Table 3.

Nasopharyngeal colonization by individual serotypes, stratified by vaccination status with regard to a 9-valent pneumococcal conjugate vaccine (PCV-9) or placebo, administered during infancy, and by HIV infection status.

The prevalence of pneumococcal colonization was less, but not significantly so, in children who had received systemic antibiotics during the previous month (38 [51.4%] of 74 vs. 158 [56.9%] of 278, respectively [P= .13]), in HIV-uninfected children (19 [39.6%] of 48 vs. 119 [53.4%] of 223, respectively [P= .08]) but not HIV-infected children (P= .84). The differences were similar when receipt of β-lactams was considered (P= .06, 0.12, and 0.93, respectively). A higher, albeit nonsignificant, prevalence of pneumococcal colonization was found in HIV-infected children receiving cotrimoxazole prophylaxis (44 [77.2%] of 57 vs. 14 [58.3%] of 24 [P= .08]) and in children with severe immunological suppression (21 [80.8%] of 26 for children with CD4 <15% vs. 3 [50.0%] of 6 children with CD4 ⩾15% [P= .15]). CDC clinical categorization (χ² test for trend, 1.31 [P= .25]) and viral load categorization (χ² test for trend, 1.94 [P= .16]) were not associated with a greater prevalence of pneumococcal colonization.

There was no difference, in the prevalence of colonization by S. aureus or H. influenzae, between PCV-9 recipients and placebo recipients, regardless of whether they were stratified by HIV status (table 2). HIV-infected children were 1.42-fold (95% CI, 1.20–1.69) (table 2) more likely than HIV-uninfected children to be colonized by H. influenzae, by at least 2 bacteria (mean RR, 1.48 [95% CI, 1.22–1.79]), and by all 3 studied bacteria (P= .003) (table 4). Children without pneumococcal colonization were more likely to be colonized by S. aureus (44.2%) than were children with pneumococcal colonization (25.0% [P= .0001]) (table 5). This inverse association also was evident among HIV-uninfected children only (44.4% vs. 21.7%, respectively; P < .0001) but not among HIV-infected children (P= .37). Among HIV-uninfected children, the inverse association between S. pneumoniae colonization and S. aureus colonization was evident in nonvaccine serogroups (colonization with S. aureus in 20 [23.0%] of the 87 children colonized by nonvaccine serogroups vs. 98 [37.0%] of the 265 children not thus colonized [P= .02]) and in vaccine serogroups (colonization with S. aureus in 15 [22.1%] of the 68 children colonized by vaccine serogroups vs. 74 [36.5%] of the 203 children not thus colonized [P= .03]).

Table 4.

Comparison of nasopharyngeal colonization by Streptococcus pneumoniae, by Staphylococcus aureus, and by Haemophilus influenzae, in HIV-infected children and HIV-uninfected children.

Table 5.

Association of nasopharyngeal colonization between Streptococcus pneumoniae, Haemophilus influenzae, and Staphylococcus aureus among children infected and not infected with HIV.

An inverse association between colonization by S. aureus and colonization by H. influenzae also was observed. Children not colonized by S. aureus were more likely to be colonized by H. influenzae (62.4%) than were those colonized by S. aureus (46.6% [P= .005]). This inverse association between S. aureus and H. influenzae colonization was also only evident among HIV-uninfected children (58.2% vs. 39.3%, respectively [P= .004]) (table 5). Conversely, children colonized by S. pneumoniae were more likely to be colonized by H. influenzae (76.5%) than were children not colonized by S. pneumoniae (32.7% [P< .0001]), and this also was true for HIV-uninfected children only (73.9% vs. 29.3%, respectively [P< .0001]) and HIV-infected children only (82.8% vs. 52.2%, respectively [P< .0001]) (table 5).

Discussion. The present data support a UK study that found no reduction in colonization by a vaccine serotype administered a few years after a PCV and a pneumococcal polysaccharide vaccine booster [2]. On the other hand, a nested study of American Indian children 1–7 years of age who had participated in a group-randomized efficacy trial reported that the children from communities randomized to receive PCV-7 (10.3%) had a lower prevalence of vaccine-serotype colonization 27 months after vaccination than did the children from control communities (17.1%) (P= .01) [3]. The differences between the 2 studies [3] might be related to (1) the possible role played by the PCV-7 booster dose administered to the children in the American Indian study and (2) the design of the studies (individuals were randomized in the UK study whereas groups were randomized in the American Indian study). In addition, the colonization levels in the American Indian children were assessed after introducing PCV-7 had been introduced into the routine immunization program—59%–80% of the children from the PCV-7—randomized communities had received ⩾3 doses of PCV-7, compared with 20% of the children from the control-arm communities. Therefore, the smaller pool of children who remained colonized by vaccine serotypes may be a function of the high community-level coverage, resulting in a change in the dynamics of vaccine-serotype community exposure and risk of colonization.

The exact mechanism whereby PCV diminishes pneumococcal colonization by vaccine serotypes is unclear. Mouse-model studies suggest that prevention of the acquisition of pneumococci may be independent of the use of antibodies [7, 8], whereas human studies suggest that it is dependent on high concentrations of serotype-specific antibodies [9, 10]. Concentrations of antibodies after a primary series of 3 doses of PCV decrease with time, to levels that are likely lower than the threshold necessary to prevent colonization, and this may explain PCV's long-term loss of effectiveness in reducing the acquisition of vaccine-serotype colonization in the absence of booster doses of PCV.

Despite the impairments of CD4⁺ T lymphocytes and humoral immunity in HIV-infected individuals, both of which may be important in the protection against pneumococcal colonization, studies have not found HIV infection to be a risk factor for pneumococcal colonization in children [11–14]. The results from the present study, however, suggest that HIV-infected children have a higher prevalence of colonization by pneumococci, especially by vaccine serotypes. Several factors may explain why, with respect to the effect that HIV status has on the prevalence of pneumococcal colonization, the results of the present study conflict with those of other studies; these factors include (1) differences in the risk of colonization and in age-related factors, (2) investigation of healthy versus ill children, (3) case and/or control selection biases in the other studies, and (4) other, unmeasured demographic factors. Although the present study has limitations because of the lack of data on other risk factors that may influence pneumococcal colonization, considering the underlying deficiency in mucosal and systemic immunity in HIV-infected children, we believe that pneumococcalcolonization differences between HIV-infected and HIVuninfected children are biologically plausible—and that this possibly contributes to the higher burden of invasive pneumococcal disease in HIV-infected children, especially that due to vaccine serotypes [15]. The number of HIV-infected children receiving highly ART in the present study were too few to determine what effect immune reconstitution might have on pneumococcal colonization.

The negative association between S. pneumoniae colonization and S. aureus colonization that previous studies had noted among healthy children [4, 5] was confirmed, among HIVuninfected children, by the present study. The exact mechanism of this negative association remains to be fully elucidated. Contributing factors include S. aureus inhibition because of hydrogen peroxide produced by colonizing pneumococci [16] as well as the development of adaptive immunity against pneumococci at an older age, after pneumococcal colonization during early childhood. A possible failure in the development of adaptive immunity against pneumococcal colonization may explain the absence of a negative association in immunocompromised, HIV-infected children. A similar observation was made among HIV-uninfected and HIV-infected children hospitalized for pneumonia [17]. Likewise, among HIV-uninfected children, but not among HIV-infected children, an inverse relationship was also observed between S. aureus colonization and H. influenzae colonization. These findings suggest that the negative associations between S. pneumoniae colonization and S. aureus colonization and between H. influenzae colonization and S. aureus colonization observed in HIV-uninfected children but not HIV-infected children probably extend beyond direct bacterial interference, with the underlying host immunological factors playing a possible role. Although the present study was able to document important trends, some of its results need further validation, because sample sizes of 1200 HIV-uninfected children and 680 HIV-infected children would be necessary to allow some of the nonsignificant findings to provide 80% power.

Potential conflicts of interest: S.A.M. and K.P.K. have received research funding and consultancy from Wyeth Vaccines and Pediatrics.

Financial support: Wyeth Vaccines and Pediatrics (support to the phase 3 vaccine efficacy trial); Comprehensive International Program of Research on AIDS in South Africa, National Institutes of Health/Division of AIDS grant 1U19A153217 to S.A.M. and P.A., for infrastructure to perform part of the carriage studies).