Effect of Vaccination on Pneumococci Isolated from the Nasopharynx of Healthy Children and the Middle Ear of Children with Otitis Media in Iceland

Vaccination with pneumococcal conjugate vaccines (PCVs) disrupts the pneumococcal population. Our aim was to determine the impact of the 10-valent PCV on the serotypes, genetic lineages, and antimicrobial susceptibility of pneumococci isolated from children in Iceland.

attending day care centers (DCCs) (2,3), and asymptomatic carriage occurs at least once prior to the age of 2 years (4). Nasopharyngeal colonization precedes pneumococcal disease, is a major factor for horizontal transmission within the community, especially among young children (5), and reflects the pneumococcal strains circulating in the community (6). Pneumococci are among the most frequent causes of bacterial AOM (7)(8)(9), which is the most common bacterial infection in children under 3 years of age (10). Pneumococcal conjugate vaccines (PCVs) interrupt the transmission of antibioticresistant pneumococci and thus decrease the burden of disease caused by antibioticresistant isolates in immunized children. Therefore, studies of the impact of these vaccines on antibiotic resistance and serotype distribution should focus both on pneumococci from nasopharyngeal carriage and from middle ears of children of similar ages with AOM (11).
Current PCVs target only a limited number of serotypes, especially those commonly causing IPD in young children (12)(13)(14). PCVs have been implemented in the infant vaccine immunization program in over 100 countries (15), which has resulted in a decrease of IPD caused by vaccine serotypes (VTs) in vaccinated children and other age groups due to herd immunity (16)(17)(18). Among healthy children, PCV implementation has also resulted in serotype replacement in carriage where VTs have been replaced with nonvaccine serotypes, meaning that the total carriage rate has remained constant (19)(20)(21). Pneumococcal carriage has been monitored and recorded in Iceland for the past 2 decades (22)(23)(24). The carriage rates have been consistent over time, ranging from 50% to 70% in healthy preschool children 1 to Ͻ7 years of age (25). The 10-valent PCV (PHiD-CV [Synflorix; GSK]), which directly targets 10 serotypes, was introduced into the national pediatric immunization program in Iceland in April 2011 in a 2-plus-1 schedule without catch-up. No other pneumococcal vaccine was previously included. The aim of this study was to assess the impact of PHiD-CV on the distribution of pneumococcal serotypes and genetic lineages among pneumococci from the nasopharynges of healthy children and the middle ears of children with otitis media and assess any changes in antimicrobial resistance rates after vaccine implementation. that increased PostVac and were the most prevalent in children 2 to Ͻ4 years of age were of serotypes 6C (n ϭ 82, 65.0/1,000 samples; P Ͻ 0.001), 15B/C (n ϭ 81, 64.2/1,000 samples; P Ͻ 0.001), and 23B (n ϭ 69, 54.7/1,000 samples; P Ͻ 0.001) ( Table S3).
MLST/CC. Among the 987 sequenced isolates, 47 CCs (35 CCs PreVac and 41 CCs PostVac) and 104 STs (66 STs PreVac and 83 STs PostVac) were detected, and 12 CCs and 43 STs were unique to nasopharyngeal isolates. The Simpson diversity index of the STs was 0.97 for both periods. A phylogenetic tree was created with the concatenated sequences of 1,066 full-length coding loci found in 99.9% of the nasopharyngeal   carriage pneumococcal genomes. The tree was annotated with CC designations and serotypes ( Fig. 1). CC439 23F/A/B was the most common CC in both study periods. Initially, 77.6% of the isolates of CC439 23F/A/B were VT serotype 23F, but PostVac, 42.0% were NVT serotype 23B; however, serotype 23B was rare prior to vaccine introduction ( Fig. 1 and Table S4). Between the two periods, the prevalence of CC433 22F increased from two isolates (1.5/1,000 samples) to 30 (20.0/1,000 samples; P Ͻ 0.001) ( Fig. 1 and Table S4), and similarly, CC1262 15B/C increased from three isolates (2.2/1,000 samples) to 18 (12.0/ 1,000 samples; P ϭ 0.032) ( Fig. 1 and Table S4).
Among the PNSP serotype 19F isolates, 92.5% belonged to CC236/271/320 19F ( Fig.  1 and Table S4). Serotype 19A PNSP increased from 4.0% to 10.4% of isolates (P ϭ 0.021) ( Table 2), and the isolates were predominantly members of CC199 19A,15B/C ( Fig.  1 and Table S4). PNSP with serotypes 6C (also MDR) and 35B were also detected.    Serotype 6C belonged to CC315 6B/C . Serotype 35B PNSP were only detected after vaccine introduction, and they were members of CC198 35B , which was not previously detected in carriage ( Table 2, Fig. 1, and Table S4). The proportion of MDR NESp isolates increased between the two study periods from 23.2% to 33.9% (P ϭ 0.025) ( Table 2), and 82.4% of MDR NESp PostVac isolates were members of CC344 NT ( Fig. 1 and Table S4). Middle ear samples from AOM. The Department of Clinical Microbiology received 6,651 ME samples during the study period. The total annual number of ME samples decreased from 966 samples in 2009 to 421 by the end of September 2017 (Table 3). Among all 6,651 samples, 994 were positive for pneumococci and 18 isolates were excluded, as they were not stored or not viable, leaving a total of 976 isolates for further analysis. The annual number of pneumococcal isolates from ME decreased from 197 (8.5/1,000 children aged 0 to Ͻ7 years) in 2009 to 44 (1.9/1,000 children) in 2016, and by the end of September 2017, the number of isolates was 23 (1.0/1,000 children) ( Table  3). The median age of the children from which the ME isolates were obtained was 1.5 years, and 69.1% (n ϭ 674) of the isolates were collected from the youngest age group (0 to Ͻ2 years of age). The average annual number of isolates in the youngest age group (0 to Ͻ2 years) decreased from 133.7 (18.1/1,000 children aged 0 to Ͻ2 years per year) PreVac to 45.5 (6.9/1,000 children aged 0 to Ͻ2 years per year) PostVac (P ϭ 0.020) (see Table S5). The genomes of 441 (50.6%) pneumococcal isolates from 2009 to 2014 were sequenced.
Serotypes. Overall, 894/976 (91.6%) pneumococcal isolates from ME samples were successfully serotyped, but 82 isolates (8.4%) were of serotypes other than those included in the mPCR scheme. Twenty-eight serotypes were detected overall: 23 PreVac and 22 PostVac. The numbers of VT pneumococci decreased significantly between the two periods (P Ͻ 0.001) ( Table 3). The numbers of VT pneumococci decreased significantly in the two younger age groups (children 0 to Ͻ2 years, P Ͻ 0.001; and 2 to Ͻ4 years of age, P ϭ 0.005), while there was no change between the periods for the oldest age group (children 4 to Ͻ7 years of age; P ϭ 0.450) ( Table S5).
Comparison of serotypes in nasopharyngeal isolates from carriage and ME isolates from children with AOM 1 to <4 years of age. Overall, 2,291 pneumococcal isolates were obtained from children 1 to Ͻ4 years of age: 1,457 isolates from nasopharyngeal samples from carriage (n ϭ 493 PreVac and n ϭ 964 PostVac) and 834 isolates from AOM ME samples (n ϭ 467 PreVac and n ϭ 367 PostVac).
The same serotypes were among the most prevalent serotypes PreVac in both sample groups, and the levels of serotype replacement PostVac were similar in both groups ( Table 5, Fig. 3). One isolate of VT serotype 23F was detected in carriage among children 1 to Ͻ4 years of age in 2017, and no VTs were detected in children with AOM within the same age group after 2016. Serotype 6A decreased slightly PostVac among children with AOM but not in carriage (P ϭ 0.050) (Fig. 3). Serotypes 6C, 15B/C, 23A, and 23B increased (Fig. 3).
VT serotype 19F was more frequently found in AOM than in carriage in children 1 to Ͻ4 years of age in both study periods: 32.1% (268/834) of ME isolates were serotype 19F compared to 7.2% (105/1,457) of carriage isolates (P Ͻ 0.001). The NVT serotype 23B was more frequently found in carriage than in AOM in children 1 to Ͻ4 years of age: 5.0% (73/1,457) of carriage isolates were serotype 23B compared to only 1.7% (14/834) of the ME isolates (P ϭ 0.006). Serotype 6C was prevalent in both carriage and AOM among children 1 to Ͻ4 years of age, but MDR was more common in ME isolates (P Ͻ 0.001). NESp isolates were more frequently found in carriage than among children with AOM in both study periods (P Ͻ 0.001) ( Table 5).

DISCUSSION
This study shows a significant reduction of VTs in nasopharyngeal carriage of healthy children and in ME samples in children Ͻ7 years old 6 years after PHiD-CV implementation in Iceland. The total number of pneumococci isolated from the nasopharynges of children remained unchanged after vaccination due to serotype replacement by NVTs; however, at the same time, the total number of ME isolates decreased significantly. The ME samples in our study were most often from children with ruptured tympanic membranes as a consequence of AOM or from children with tympanic tubes and discharge from the middle ear as a consequence of inflammation in the middle ear. Consequently, we postulate that the decrease in the number of samples from the ME most likely reflects a decrease in the burden of AOM (52). Certain serotypes are associated with nasopharyngeal carriage in healthy children, while others are more prone to cause disease (53,54); however, the NVTs that have replaced the VTs following routine vaccination may, to a great extent, express a low invasive disease potential (55).
It was confirmed in our study that the distribution of serotypes and genetic lineages in nasopharyngeal carriage in children reflected those identified from the discharges from ears of children with AOM (56). Furthermore, the diversity of the pneumococcal STs did not change between the periods.
The same serotypes and genetic lineages were often found in nasopharyngeal and ME samples in children 1 to Ͻ4 years of age, both PreVac and PostVac but often in different proportions. There was a slight but significant decrease in serotype 6A among children with AOM, which has been described in IPD among all ages following PHiD-CV vaccination (57). Serotypes 6C, 23A, and 23B were not affected by the vaccination, as these serotypes increased significantly in both sample groups, which has also been documented elsewhere (21,(58)(59)(60)(61)(62).
Serotype 19F was particularly "otogenic," as it was four times more prevalent in ME samples than in nasopharyngeal samples. The main serotype 19F in CC236/270/320 19F harbors genes for pili (both PI-1 and PI-2), which likely contributed to the adherence to the mucosa of the middle ear (63). Serotype 6B had a preference for nasopharyngeal colonization PreVac, while the same was true for serotype 23B PostVac. This finding is in concordance with other reports where serotype 23B was associated with persistent carriage and posed a lower risk for IPD (55,64).
The NESp isolates were largely associated with carriage in both study periods but extremely rare in AOM. However, while the prevalence of NESp did not change significantly after vaccination in carriage, MDR among NESp isolates increased significantly.
In our study, only one VT isolate was detected between 2014 and 2016 in ME samples among children 1 to Ͻ4 years of age, whereas 11 VT isolates were detected between 2014 and 2017 in the nasopharynges of healthy children of the same ages. This indicates that PHiD-CV may be more effective in preventing AOM (measured as fewer positive ME samples) than in preventing nasopharyngeal carriage. This is important, as AOM caused by pneumococci is more severe than that caused by other common pathogens (65,66).
The increase of the MDR serotype 6C isolates PostVac in both nasopharynx and ME samples is of concern. Other researchers have also described an increase in serotype 6C following the introduction of PCV7, especially in non-IPD among children (67). We rarely detected isolates of serotype 6C PreVac; however, in a recent study, serotype 6C was only detected among vaccinated children with non-IPD after PHiD-CV implementation (68). Serotype 6C isolates detected PostVac most often belonged to CC315 6B and ST386 6C (a double locus variant [DLV] of PMEN Poland 6B -20). The MDR ST386 6C was detected in Spain 6 years after PCV7 implementation (69), and other countries have also reported the emergence of this lineage in nasopharyngeal carriage and IPD following the implementation of PHiD-CV (62). This lineage might be derived from a capsular switch from serotype 6B to serotype 6C, as has previously been reported by our group (36).
The majority of serotype 23B isolates belonged to CC439/ST439 23B (single locus variant [SLV] of PMEN Tennessee 23F -4); however, this lineage was only detected PostVac in both sample groups in our study. CC439/ST439 23B was present, although uncommon, in Germany prior to the implementation of vaccinations but increased after vaccine implementation. This lineage has also been documented in other countries worldwide (70).
One ME isolate of serotype 23B belonged to CC156/162 and ST162 (an SLV of PMEN Spain 9V -3). This was the only isolate of this lineage detected PostVac in ME isolates. Serotype switch variants of the related ST156 expressing serotypes 9V, 9A, 14, 19F, and 11A have also been reported (71). Pneumococci have the ability to change their capsular serotype by exchanging the capsular locus genes (72). The expansion of preexisting lineages and their variants, such as CC315 6B/C and CC439 23F/B , may be more likely following PCV vaccination than the emergence of new lineages (70).
Iceland offers a unique opportunity for researching vaccine effects for several reasons. The reference laboratory at the Department of Clinical Microbiology, Landspitali University Hospital, serves approximately 85% of the country for pneumococcuspositive samples and stores all isolates (at Ϫ80°C). Furthermore, carriage studies have been conducted within the same DCCs using the same methodology throughout the study period. PCVs were not part of the routine infant immunization program before PHiD-CV implementation in 2011. Since then, vaccine acceptance and the uptake of PCVs have been high (73,74). In our study, we analyzed a large number of pneumococcal isolates representative of the Icelandic population. Furthermore, one third of the pneumococcal isolates were subjected to WGS, which gives a good overview of the composition of genetic lineages in the country. However, fluctuations in serotypes and genotypes are known among pneumococci, even without the selective pressure of PCVs (75,76).
In conclusion, PHiD-CV implementation eliminated VTs in the MEs of children with otitis media within 5 years. The carriage rate of pneumococci in healthy children remained constant between the periods due to serotype replacement of NVTs, but the total number of ME isolates decreased significantly PostVac. Serotype 23B and NESp had a preference for nasopharyngeal carriage among children 1 to Ͻ4 years of age. Multidrug resistance among serotype 6C was more common in ME samples among children 1 to Ͻ4 years of age than in nasopharyngeal carriage samples PostVac.

ACKNOWLEDGMENTS
This was an investigator-initiated study funded by GlaxoSmithKline Biologicals SA. GlaxoSmithKline Biologicals SA was provided the opportunity to review a draft version of the manuscript, but the authors are solely responsible for the final content and interpretation. The authors received no financial support or other form of compensation related to the development of the manuscript.
We thank the staff at the Department of Clinical Microbiology for collecting the isolates from ME samples. We also thank the teachers and staff at the DCCs and the children participating in the carriage studies and their parents. We also thank other members of the VIce study group.