Background. The impact of the 7-valent pneumococcal conjugate vaccine (PCV7 [Prevnar]) on penicillin-nonsusceptibleStreptococcus pneumoniae (PNSP) recovered from children with acute otitis media (AOM) is unclear.
Methods. At 5 hospitals, 505 pneumococcal isolates were collected from children with AOM between 1 January 1999 and 31 December 2002. Molecular subtyping was performed on 158 isolates.
Results. Overall, the percentage of AOM cases due to non-PCV7 serogroups (including serotype 3) increased over time (from 12% in 1999 to 32% in 2002;P < .01) and according to the number of PCV7 doses received (18% [⩽1 dose] vs. 35% [2–4 doses];P < .01). The percentage of cases due to vaccine-related serotypes (including serotype 19A) increased according to the number of PCV7 doses received (10% [⩽1 dose] vs. 19% [2–4 doses];P = .05) but not over time, whereas the percentage of cases due to serotype 19F remained unchanged both over time and according to the number of PCV7 doses received. The frequency of penicillin nonsusceptibility among PCV7 serotypes (range, 65%–75%) and non-PCV7 serogroups (range, 11%–27%) did not significantly change overall. Although no change was detected among isolates collected from children with spontaneous drainage, the percentage of pneumococci recovered at the time of myringotomy and/or tympanostomy tube placement that were nonresistant to penicillin decreased over time (from 73% in 1999 to 53% in 2002;P = .03). All of the serotype 3 strains were genetically related, whereas 88% of the isolates that were either serotype 19F or serotype 23F were related to 1 of 3 international clones.
Conclusions. Among children with AOM, the proportion of cases due to non-PCV7 serogroups increased, vaccine-related serotypes increased, and serotype 19F remained unchanged. Although a decrease in the proportion of cases due to PNSP occurred among children who required myringotomy and/or tympanostomy tube placement, the proportion of PNSP remained unchanged overall and among children with spontaneous drainage. Because future trends in the susceptibility patterns of pneumococcal isolates recovered from children with AOM are not easy to predict, continued surveillance is essential.
The 7-valent pneumococcal conjugate vaccine (PCV7 [Prevnar; Wyeth Lederle Vaccines]) provides only a modest amount of protection against acute otitis media (AOM) [1–3]. When administered to infants, it has substantially reduced cases of recurrent AOM and the necessity for placement of tympanostomy tubes in healthy infants [2, 4]. In contrast, this vaccine has not significantly decreased episodes of AOM when administered to children who have a history of recurrent AOM . Moreover, serotype replacement has been observed in randomized, controlled trials  and observational studies  comparing the antimicrobial susceptibility of middle ear isolates ofStreptococcuspneumoniae recovered during the prevaccine era with the susceptibility of such isolates recovered during the postvaccine-licensure period.
The incidences of invasive pneumococcal disease and penicillin-nonsusceptibleS. pneumoniae (PNSP) infection have substantially decreased since licensure of PCV7 [7–9]. Some studies suggest that the proportion of invasive cases due to PNSP stabilized in 2001  before decreasing in 2002 [7, 10]. However, the proportion of cases due to PNSP strains has remained unchanged among children with nasopharyngeal carriage . The impact of PCV7, if any, on the proportion of PNSP isolates causing AOM is unknown. Because tympanocentesis is not performed routinely when children present to their primary care physicians for evaluation of AOM, the effect of PCV7 on the incidence of ear infections due toS. pneumoniae must be extrapolated from clinical and observational studies. For example, middle ear isolates from children with AOM and spontaneous ear drainage may be surrogate markers for typical AOM, because there is no known correlation between virulence and spontaneous perforation. Furthermore, the recovery of middle ear isolates ofS. pneumoniae from children at the time of tympanostomy tube placement may serve as a surrogate for recurrent AOM [12, 13].
The aim of this investigation was to determine whether the proportions of AOM cases due to PNSP strains, PCV7 serotypes, vaccine-related pneumococcal serotypes, and non-PCV7 serogroups among children changed between 1999 and 2002. In addition to examining the proportions of all isolates recovered from children, we performed separate analyses for children with spontaneous ear drainage and for children who had isolates ofS. pneumoniae recovered at the time of myringotomy and/or tympanostomy tube placement to determine whether an association between clinical presentation and PNSP was present. Molecular analyses of serotype 3, 6A, 6B, 19A, 19F, and 23F isolates were performed for patients at 1 center to correlate the genotypes of the isolates causing AOM with the epidemiologic trends.
Patients and Methods
Patient enrollment, isolate collection, and study definitions. Middle ear isolates were obtained from children with AOM treated at the following institutions participating in the US Pediatric Multicenter Pneumococcal Surveillance Study Group: Children's Hospital of Pittsburgh at the University of Pittsburgh Medical Center, Texas Children's Hospital and Baylor College of Medicine (Houston), Children's Hospital San Diego, Children's Hospital of Los Angeles and the University of Southern California School of Medicine, and Arkansas Children's Hospital and University of Arkansas for Medical Sciences (Little Rock), as described elsewhere . Isolates were collected from 1 January 1999 through 31 December 2002 at all sites, except for Children's Hospital San Diego, where isolates were not collected in 2002. This period was chosen to represent strains that were causing AOM before and after licensure of PCV7 in 2000.
One vaccine dose was defined as the receipt of PCV7 at least 1 month before infection. Recent antibiotic use was defined as antibiotic use ⩽30 days before infection. Because tympanocentesis is often performed in cases of clinical failure, recurrent infection was defined as the recovery of 2 isolates ofS. pneumoniae from the same child at least 14 days apart. Samples of middle ear fluid were obtained from spontaneous drainage during an episode of AOM, during myringotomy and/or tympanostomy tube placement, or during tympanocentesis performed for children who did not respond to clinical treatment or who received a diagnosis of AOM during a clinical trial.
Antibiotic susceptibility and serogrouping. Testing for penicillin susceptibility (by the NCCLS microbroth dilution method) and serotyping (with the use of reagents from Statens Serum Institut [Copenhagen, Denmark]) were performed. PCV7 includes serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F. Vaccine-related serotypes include serotypes 6A and 19A. Sixteen isolates from PCV7 serogroups were not available for serotyping and were excluded from the serotype distribution analyses. Isolates that were intermediately resistant or resistant to penicillin were classified as penicillin nonsusceptible, with penicillin MICs of 0.1–1 µg/mL and ⩾2 µg/mL, respectively .
PFGE and multilocus sequence typing (MLST). PFGE was performed on 162 (99.4%) of 163 serogroup 3, 6, 19, and 23 isolates from patients at the Children's Hospital of Pittsburgh [16, 17]. Dendrograms were created in Bionumerics (Applied Maths) by means of the unweighted pair group method using arithmetic averages, the Dice coefficient, and a position tolerance of 1.5%. MLST was performed on 85 (52.5%) of 162 isolates as described elsewhere . Alleles were assigned through the Multi Locus Sequencing Typing Web site (available at: http://www.mlst.net). Isolates were compared to the first 25 international clones: Spain23F-1, Spain6B-2, Spain9V-3, Tennessee23F-4, Spain14-5, Hungary19A-6, South Africa19A-7, South Africa6B-8, England14-9, Slovakia14-10, Slovakia19A-11, Finland6B-12, South Africa19A-13, Taiwan19F-14, Taiwan23F-15, Poland23F-16, Maryland6B-17, Tennessee14-18, Columbia5-19, Poland6B-20, Portugal19F-21, Greece6B-22, North Carolina6A-23, Utah35B-24, and Sweden15A-25 [18–25].
Analysis. All isolates were initially classified by PFGE. MLST was performed on the spectrum of PFGE patterns, including the outliers of each PFGE-based cluster. Isolates that were clustered according to PFGE patterns and/or that had at least 5 identical alleles according to MLST were classified into sequence type (ST) complexes. Isolates within a ST complex were designated as clones in the statistical analyses. With one exception, the PFGE finding correlated with the targeted MLST. The exception, a serotype 23F isolate, had a PFGE pattern that differed by 6 bands from that of the Tennessee23F-4 clone but had only 3 of 7 alleles in common. In this case, the isolate was classified by PFGE rather than by both PFGE and MLST.
Statistical analyses were generated with SAS, version 8.02 (SAS Institute). Exact tests were calculated by StatXact, version 6.0 (Cytel Software), and Epi Info, version 6.0 (Centers for Disease Control and Prevention). We performed analyses of temporal trends and examined the number of PCV7 doses for serogroups found in at least 20 isolates across the 5 sites. Serotypes 3, 6A, 6B, 9V, 14, 19A, 19F, and 23F were included. For analyses that were based on clinical presentation, only significant changes in the serotype distribution of PNSP isolates are described.
Overall analysis. A total of 505 cases of AOM due toS. pneumoniae were included in the analysis. The number of cases per study site were as follows: Pittsburgh, 257 cases; Houston, 154; San Diego, 57; Little Rock, 19; and Los Angeles, 18. The total number of available cases decreased between 1999 and 2002 across the 5 sites (table 1). The percentage of cases due to non-PCV7 serogroups (including serotype 3) significantly increased from 12% in 1999 (22 of 182 cases) to 32% in 2002 (26 of 82 cases) (P < .01). Likewise, the percentage of cases due to PCV7 serogroups decreased from 76% in 1999 (139 of 182 cases) to 52% in 2002 (43 of 82 cases) (P < .01). Among PCV7 serotypes, the proportions of cases due to serotypes 6B, 14, and 23F significantly decreased from 1999 to 2002. In contrast, the proportions of cases due to serotype 19F and to PCV7-related serotypes remained unchanged over time.
PCV7 dose analyses revealed results that were similar to those of the temporal analyses, with 2 exceptions. The proportion of cases due to serotype 23F was not significantly different with respect to the number of PCV7 doses received. Moreover, among children who received ⩽1 dose of PCV7, 43 (10%) of 435 were infected with a vaccine-related serotype, compared with 12 (19%) of 63 who received 2–4 doses of PCV7 (P = .05; table 1). The serogroup distribution of non-PCV7 isolates was as follows: serogroup 3 was recovered from 26 children (5%); serogroup NT was recovered from 18 (4%); serogroup 11 was recovered from 13 (3%); serogroup 15 was recovered from 12 (2%); serogroups 29, 35, and 42 were recovered from 12 (2%); serogroup 16 was recovered from 5 (1%); serogroup 22 was recovered from 3 (1%); serogroup 10 was recovered from 3 (1%); serogroup 7 was recovered from 2 (<1%), and serogroups 1, 20, 29, 31, 33, 37, and 38 were recovered from 1 (<1%) each.
Overall, 59% of the isolates were PNSP, 22% and 37% of which were intermediately resistant and resistant, respectively, to penicillin (table 2). The proportions of isolates that were nonresistant to penicillin, intermediately resistant to penicillin, and resistant to penicillin remained unchanged over time and did not vary according to the number of PCV7 doses received. Per year, the proportion of strains that were penicillin nonsusceptible ranged from 66% to 88% among serogroups included in PCV7 and from 12% to 27% among serogroups not included in PCV7 (table 2). Except for isolates from serogroup 15, serogroups 29, 35, and 42, and serogroup NT, all isolates from non-PCV7 serogroups were susceptible to penicillin.
Analysis of isolates obtained during spontaneous ear drainage. For 236 (47%) of 505 children with spontaneous ear drainage, the percentage of AOM cases due to non-PCV7 serogroups increased from 13% in 1999 (8 of 62 cases) to 36% in 2002 (16 of 45 cases) (P < .01), but no difference between the periods was detected by PCV7 dose analyses. From 1999 to 2002, the proportions of cases due to serotypes 3 and 19F significantly increased, whereas the proportion of cases due to PCV7 serotypes, including serotype 23F, decreased. A higher proportion of cases was due to serotype 3 and a lower proportion was due to serotype 6B among children who received at least 2 doses of PCV7, compared with children who received fewer doses (data not shown). The percentage of children with PNSP did not change significantly over time (range, 43%–60%) or by PCV7 dose (range, 50%–56%). However, among PCV7 serotypes, the percentage of PNSP isolates increased significantly over time (from 60% in 1999 [25 of 42 isolates] to 100% in 2002 [24 of 24 isolates];P < .01) and by PCV7 dose (from 62% for ⩽1 dose [78 of 125 isolates] to 94% for 2–4 doses [17 of 18 isolates];P ⩽ .02).
Analysis of isolates obtained during myringotomy and/or tympanostomy tube placement. For 186 (37%) of 505 patients who had pneumococci recovered at the time of myringotomy and/or tympanostomy tube placement, the percentage of isolates from non-PCV7 serogroups substantially increased over time, from 12% in 1999 (10 of 83 isolates) to 30% in 2002 (9 of 30 isolates) (P = .01). Likewise, 8 (35%) of 23 patients who received at least 2 doses of PCV7 were infected with an isolate from a non-PCV7 serogroup, compared with 22 (14%) of 161 who received fewer doses of PCV7 (P = .03). From 1999 to 2002, for patients who received at least 2 doses of PCV7, the proportions of cases due to PCV7 serotypes and serotype 14 decreased and the proportion due to vaccine-related serotypes increased, compared with patients who received fewer doses of PCV7 (data not shown). The percentage of PNSP isolates recovered during myringotomy and/or at the time of tympanostomy tube placement decreased from 73% in 1999 (61 of 83 isolates) to 53% in 2002 (16 of 53 isolates) (P = .03). Significant variation in the proportion of PNSP isolates with respect to the number of PCV7 doses received was not detected.
Molecular analysis of strains from one site. In general, data from Children's Hospital of Pittsburgh mirrored results of the overall temporal analyses (table 3). The increase in the proportion of cases due to serotype 3 strains was clonal. Among 83 serotype 19F and 23F strains, 73 (88%) were related to 1 of 3 international clones. The proportion of cases due to strains related to the Taiwan19F-14 and Tennessee23F-4 clones dramatically changed over time.
This multicenter, observational study of children with AOM due toS. pneumoniae demonstrates that the proportion of AOM cases caused by PNSP has remained unchanged during 1999–2002. These data also confirm results of an earlier study demonstrating that the proportion of cases due to serogroups not included in PCV7 has increased over time and among children who received at least 2 doses of PCV7, compared with children who received ⩽1 dose . The proportion of cases due to vaccine-related serotypes did not decrease over time. Infections due to vaccine-related serotypes were more common among children who received at least 2 doses of PCV7 than among children who received fewer doses. Despite a decrease in the proportion of cases due to PCV7 serotypes, overall and subgroup analyses revealed that the proportion of cases due to serotype 19F isolates did not decrease. The molecular analyses indicate that this increase was largely due to isolates related to 2 international clones; isolates related to the Taiwan19F-14 clone accounted for most of the isolates. Serologic data indicate that the concentration of antibodies against serotype 19F does not correlate with the risk of AOM due to serotype 19 isolates . Taken together, these data suggest that PCV7 is less protective against serotype 19F isolates than are other serotypes included in PCV7 [7, 9].
Overall, serotype 3 accounted for an increasing proportion of AOM cases, as revealed by the temporal analyses and the PCV7 dose analyses. Among the cases from Children's Hospital of Pittsburgh that we analyzed, the serotype 3 isolates were clonal but did not arise through capsular transformation with any of the first 25 international clones. Serotype 3 isolates accounted for 20% of the pneumococcal isolates colonizing the nasopharynx of adults in the PCV7 era , suggesting that this serotype may cause more infections in both children and adults in the future.
Separate analyses were performed for children with spontaneous ear drainage and for children who hadS. pneumoniae isolated at the time of tympanostomy tube placement. Cases associated with spontaneous drainage most likely reflect the trends occurring among children with typical AOM, whereas cases requiring tympanostomy tube placement most likely reflect the trends among children with persistent or recurrent AOM. Regardless of the source of the isolate, the proportion of cases due to serogroups not included in PCV7 significantly increased between 1999 and 2002. Among children with spontaneous ear drainage, the percentage of pneumococci that were nonsusceptible to penicillin did not change between 1999 and 2002, despite an increase in the proportion of PNSP among serotypes included in PCV7. In contrast, among the children who hadS. pneumoniae isolates detected at the time of myringotomy and/or tympanostomy tube placement, the proportion of pneumococci that were nonsusceptible to penicillin decreased.
This study has some limitations. First, because the data do not include all cases of AOM, the impact of PCV7 on the incidence of AOM cannot be assessed. Second, a correlation between the temporal analyses and PCV7 dose analyses was not always present. This may be due to the use of the PCV7 dose as a dichotomous rather than continuous variable. Alternatively, the discrepancy may be due to information bias associated with incomplete documentation. Because of the frequent vaccine shortages that have occurred since licensure in 2000 and because of suboptimal vaccination rates , tabulation of the number of PCV7 doses received on the basis of age was not a reasonable alternative.
In summary, although the proportion of AOM cases due to pneumococci that were nonsusceptible to penicillin did not change overall, a decrease was noted among children who had undergone myringotomy and/or tympanostomy tube placement. The proportion of cases due to serogroups not included in the PCV7 vaccine, including a clonal increase in serotype 3, increased over time and among those who received at least 2 doses of PCV7, compared with those who received fewer doses. Despite a decrease in the proportion of cases due to PCV7 serotypes, the proportions of cases due to serotype 19F and vaccine-related serotypes did not decrease over time. These data indicate that PCV7 has not been associated with a marked reduction in the proportion of AOM cases due to PNSP, although an effect may be present among children who have undergone myringotomy and/or tympanostomy tube placement. Among cases of AOM, PCV7 appears to provide less protection against strains from vaccine-related serotypes and serotype 19F, compared with the other PCV7 serotypes. Clinical trials focusing on typical AOM and recurrent AOM are needed to validate our results. Because future trends in the penicillin susceptibility of isolates ofS. pneumoniae recovered from children with AOM are not easy to predict, continued surveillance is essential.
We gratefully acknowledge Michelene Ortenzo, Diana Kearney, Karen Barbardora, Andrea Forbes, Bev Petrites, Linda Lamberth, and Nancy C. Tucker, for data collection and laboratory support.
Financial support. National Institutes of Health (grant K23-AI01788-01 to M.C.M.), Bristol-Myers Squibb (to M.C.M. and E.R.W.), and Roche (to E.O.M., E.R.W., S.L.K., G.E.S., and J.S.B.)
Potential conflicts of interest. G.E.S. has received research funding from Aventis; E.O.M., E.R.W., S.L.K., G.E.S., and J.S.B. have received research funding from Roche; and M.C.M and E.R.W. have received research funding from Bristol-Myers Squibb. All other authors: no conflicts.