Carriage of Streptococcus Pneumoniae In Unvaccinated Toddlers At The Time of Pneumococcal Conjugate Vaccine Implementation Into The National Immunization Program In Poland

We investigated pneumococcal carriage among unvaccinated children under ve years of age at the time of conjugate polysaccharide vaccine (PCV) introduction into the national immunization program (NIP). Paired nasopharyngeal swab (NPS) and saliva samples collected between 2016 and 2020 from n=394 children were tested with conventional culture and using qPCR. The carriage rate detected by culture was 25.4% (97 of 394), by qPCR 39.1% (155 of 394), and 40.1% (158 of 394) overall. The risk of carriage was signicantly elevated among day care center attendees, and during autumn/winter months. Among strains cultured, the most common serotypes were: 23A, 6B, 15BC, 10A, 11A. The coverage of PCV10 and PCV13 was 23.2% (23 of 99) and 26.3% (26 of 99), respectively. Application of qPCR lead to detection of 168 serotype carriage events, with serogroups 15, 6, 9 and serotype 23A most commonly detected. Although the highest number of carriers was identied by testing NPS with qPCR, saliva signicantly contributed to the overall number of detected carriers. Co-carriage of multiple serotypes was detected in 25.3% (40 of 158) of carriers. Results of this study represent a baseline for the future surveillance of effects of pneumococcal vaccines in NIP in Poland.


Introduction
Streptococcus pneumoniae is the common cause of invasive bacterial disease [1,2]. Incidence of invasive pneumococcal disease (IPD) is highest among infants and toddlers and in older adults [1]. Despite available vaccines, in 2015 pneumococcus was responsible globally for approximately 300 000 deaths of children under 5 years of age [2]. IPD is manifested by meningitis, sepsis and/or bacteremic pneumonia. S. pneumoniae also causes milder infections manifested across all ages as sinusitis or non-bacteremic pneumonia, and in children as acute otitis media.
The primary virulence factor of pneumococci is the polysaccharide capsule [3] and currently available pneumococcal vaccines are all based on capsular polysaccharides as antigen. While there have been almost 100 capsular types (serotypes) described [4], marketed vaccines target only a subset of ten to twenty-three serotypes (24 in total). Pneumococcal conjugate vaccines (PCV) recommended for children [2], target ten (PCV10) and thirteen (PCV13) serotypes common in paediatric disease prior to PCVs introduction. PCV10 and PCV13 have been commercially available in Poland since 2009 and 2010, respectively. In 2017, PCV10 was introduced into the National Immunization Program (NIP) for all children born after 31st of December 2016, and PCV10 was chosen as the refunded vaccine in all consecutive annual NIP tenders. Children are vaccinated at 2, 4, and 13-15 month of life (2 + 1 schedule). However, it is estimated that a quarter to a third of infants in Poland is vaccinated with PCV13 outside NIP [5].
Three years after PCV10 implementation into NIP, there was a signi cant decrease of IPD cases caused in Poland by PCV10 vaccine serotypes (VT) in children under 2 years of age (57% in years 2014-2016 vs. 31% in years 2017-2019) [6]. There was also a signi cant decline from 35-28% in PCV10-VTs IPDs in persons ≥ 65 years old. Similar herd effects were earlier observed in other countries [7][8][9]. Indirect effects are attributed to PCVs preventing VT strains carriage acquisition in vaccinees [10,11]. Since children are the main reservoir and the main transmitters of pneumococci, infants' vaccination with PCVs may have impact on carriage and disease in unvaccinated individuals in the same population [3,12,2]. Consequently, effects of PCVs can be monitored not only via surveillance of disease but also of carriage [13].
Since 1997 the National Reference Centre for Bacterial Meningitis (NRCBM) collects isolates causing IPD from the whole country and also conducts molecular diagnostics of IPD cases. Although surveillance of IPD in Poland is well established, the data on S. pneumoniae and pneumococcal serotypes carriage is limited. Over the past twenty years there have been only two studies published on pneumococcal carriage conducted in Poland [14][15][16][17]. To ll the gap, we investigated the pneumococcal carriage in unvaccinated children under ve years of age. Our goal was to map the carriage of S. pneumoniae serotypes in paediatric population before the nationwide immunization of infants with PCV10 may result in substantial indirect effects. With this, we were aiming to establish a baseline for the future surveillance studies.
The gold standard for pneumococcal carriage detection is isolation of live S. pneumoniae from a culture of a deep trans-nasal nasopharyngeal swab (NPS) [18,19]. There is evidence that other samples, oral uids in particular, may substitute for NPS [20,21]. Compared to NPS, saliva is much easier to collect, can be even self-collected, and, except for the youngest children, it does not require a designated collection kit. It has been also reported that application of molecular methods allows higher sensitivity of S. pneumoniae and pneumococcal serotypes carriage detection [22][23][24]. Hence, our second goal was to compare results of carriage detection by testing saliva versus NPS, and using molecular methods versus conventional diagnostic culture in order to develop a procedure tailored to our future studies [25][26][27]. To our knowledge, there are no published studies comparing saliva testing with the gold standard in toddlers.

Results
Altogether 405 children have been enrolled in the study. Of these, nine children were excluded from further analysis for either being too young (n = 4), too old (n = 4), or being diagnosed with lower respiratory tract infection on enrolment day (n = 1). Two children were enrolled twice in the study and results of the rst sampling were the only included. We report results for paired NPS and saliva collected once from 394 children.

Study population
Frequency of inclusions declined over the study years ( Table 1) with half of children enrolled by the 14th month (September 2017) of 44-months long project. The number of 12-23-months old children (n = 147 of 394) was signi cantly higher compare with any other age group (Fisher Exact, p < 0.001), and the number of 24-35 months old (n = 102) was signi cantly higher compare to 36-47 months and 48-59 months olds (n = 71 and n = 74, respectively, p < 0.01).  (Fig. 1b). With 94 of these 97 children positive in NPS and six in saliva, NPS was far superior to saliva in the sensitivity of detecting pneumococcal carriage by isolation of live S. pneumoniae from a child (McNemar's, p < 0.000001). All 394 saliva samples yielded a colony growth on GENT-agar and all these plates were harvested whereas 68 (17.3%) of 394 NPS cultures were negative for any growth. We considered these 68 NPS to be negative for S. pneumoniae also by molecular method.
Samples from 155 children (39.1% of 394) have been identi ed as positive for pneumococcus with qPCRs ( Fig. 1b). It included all samples from which S. pneumoniae has been cultured except for three NPS samples positive by culture for non-typeable (NT) pneumococci (Fig. 1a), and a single saliva sample from which serotype 24F strain has been cultured (Fig. 1c). Similar to diagnostic culture, with molecular methods the number of positive results was higher for NPS compared to saliva (121 or 30.7% versus 93 or 23.6%; McNemar's, p < 0.01) (Fig. 1b) Table 1). RR was also elevated during autumn and winter months. Being a sibling was associated with a higher risk of carriage in the study (RR1.52(1.04-2.89), p < 0.05), but only when S. pneumoniae was detected by culture of NPS and was driven primarily by an effect observed in 12-23 month-olds (RR2.40(1.28-4.51), p < 0.001) and not in other age groups (24-35 months old, p = 0.70; 36-47 months old, p = 0.17; 48-59 months old, p = 0.66). After correcting for DCC attendance an impact of the siblings has become insigni cant (RR1.02(0.76-1.38), p = 0.89). RR was higher in children from households without a smoker (Chi-square, p < 0.05), but only when carriage was detected by culture of NPS. There was no effect of age or gender on prevalence of carriage detected by a particular method or overall (Table 1).

Serotype carriage
Serotypes of strains cultured from children and detected with qPCR in culture-enriched samples in the study are all listed in Table 2. Altogether 99 strains have been isolated from 97 children, as S. pneumoniae strains of two different serotypes have been cultured from NPS collected from two individuals. Ninety-three of these 99 isolates represented 26 different serotypes and the remaining six were classi ed as non-typeable pneumococci. The most common serotype among cultured S. pneumoniae strains were 23A and 6B isolated from 10 children each, followed by 15BC, 10A and 11A isolated from seven children each, and by serotypes 23B and 35F isolated from six children each. Strains of these seven serotypes constituted 53.5% of 99 strains cultured in the study. The coverage of PCV10 and PCV13 was 23.2% (23 of 99) and 26.3% (26 of 99), respectively.    Among 158 children classi ed as carriers of S. pneumoniae by any method, the most common serotype/serogroup detected with qPCRs were: serogroup 15 (n = 24 or 15.2% of 158 children), serogroup 6 (n = 22 or 13.9%), serogroup 9 and serotype 23A (n = 17 or 10.8%, each). For NPS, numbers of samples positive for serotype or serogroup by culture correlated strongly with the number of samples positive for the same serotypes by qPCR (Spearman's rho = 0.855, p < 0.0001) (Fig. 2a). There was also correlation between the number of serotype carriers detected with qPCR in NPS and in saliva (rho = 0.667, p < 0.005) (Fig. 2b), as well as between numbers of serotype-carriers detected with qPCR in NPS or in saliva versus overall cultured in the study (rho = 0.771, p < 0.0001) (Fig. 2c).
Although the number of VT strains cultured from children attending DCC (n = 17 of PCV10-VT and n = 20 of PCV13-VT) was signi cantly higher compared with the number of cultured children staying home (n = 6 of PCV10-VT and PCV13-VT, Chi-square, p = 0.034 and p = 0.01, for difference in prevalence of PCV10-VTs and PCV13-VTs, respectively), there were no differences in fractions of PCV10-VT strains among all cultured from DCC attendees compared with children staying home (17 of 71 versus 6 of 28, Chi-square p = 0.79), nor in fractions of PCV13-VT strains (20 of 71 for PCV10 versus 6 of 28, Chi-square p = 0.49). None of other demographic or environmental factors were associated with differences in serotype carriage.

Co-carriage of multiple serotypes
Presence of two or more serotypes was detected in 40 (25.3%) of 158 children identi ed as carriers by either culture or using piaB and lytA qPCRs (   [20,33]. NPS is the specimen recommended by WHO in pneumococcal carriage detection in children [19] and it has been reported that NPS is a more valuable material than OPS [35]. In our study, the culture of NPS was far superior to culture of saliva and also, 39.6% of carriers (61 of 154) identi ed by qPCR were detected by NPS only. However, with 21.5% (34 of 158) of carriers detected exclusively in saliva, testing oral uids substantially increased the number of carriers detected. In line with this nding, Korona-Glowniak et al. [16,17] also reported that testing OPS along NPS signi cantly increased the number of carriers detected, and that there was no difference between the number of carriers detected by culturing NPS compared with OPS [16,17]. Therefore, the best carriage detection might be achieved by testing from each individual multiple specimens, e.g. NPS, OPS and saliva, or a combination of any two of these.
Molecular methods appeared to be superior to conventional culture in detecting co-carriage of multiple serotypes in this study (2.1% in culture versus 26.6% in qPCR). Wyllie et al. [33] and Kandasamy et al. [36] obtained similar levels of multiple serotype carriage using molecular methods. The higher sensitivity of any minority serotype detection in multiserotype carriage allowed for a more detailed analysis of the occurrence of serotypes. Since available qPCR assays did not cover all serotypes, and not always distinguished serotypes within a serogroup, the number of multi-serotype carriers still might be understated.
Among our study limitations was a lack of molecular assays that would detect carriage of every circulating serotype. For example, strains of serotypes 24F, 28F, 35A, 35F, and 38 have been cultured from children, yet none of these serotypes were targeted with qPCR. Another limitation was low resolution of certain qPCRs not discriminating between serotypes within a serogroup, with 10 out of 27 qPCR assays targeting more than one serotype. A limitation was also the low sensitivity of conventional diagnostic culture. It concerns both sample types, but due to very rich bacterial growth, including many non-pneumococcal α-hemolytic colonies, it was particularly di cult to isolate live S. pneumoniae from saliva. With large numbers of serotype-carriage events detected exclusively with qPCR, and in the light of reports on non-pneumococcal streptococci expressing the pneumococcal capsular polysaccharides [28, 37], we paid particular attention to the speci city of assays. We addressed it by testing for serotype samples negative for S. pneumoniae and excluding the results of assays that generated a positive result (serotype 4 and serotype 5 speci c qPCRs). Nevertheless, we can't exclude that some of the results represent carriage of confounded non-pneumococcal bacteria detected with qPCRs. For example, when applied to oropharyngeal and saliva samples from adults, a diminished speci city of serogroup 9-speci c assay has been reported [26, 28] and serogroup 9 was the clear outlier when culture data was compared with qPCR results in our study (Table 2 and Fig. 2a and 2c). However, since we did not observe positivity in this assay among samples negative for S. pneumoniae, nor was there a difference between the number of NPS and saliva samples positive for this serogroup by qPCR, and we are not aware of any reports on the assay's poor speci city in NPS from children, we consider results for serogroup 9 as reliable.
In summary, pneumococcal carriage rate detected with WHO's recommended method in Polish children was lower compared with studies conducted prior to the introduction of commercial PCVs in the country, yet we attribute it to differences in set-ups of studies rather than the effect of PCVs. On the other hand, the decline in prevalence of PCV10-VTs and PCV13-VTs carriage compared with the pre-PCV period suggests strong herd effects of commercial vaccination independent of NIP in Poland. According to the results obtained in our study, NPS was a more valuable material in carriage detection in children and qPCR was the more sensitive method in pneumococcal and pneumococcal serotype carriage detection. Also, information about carriage rate and serotype distribution among unvaccinated Polish children gained during this study can be used as a baseline in future carriage projects. The knowledge concerning the methods used in pneumococcal carriage detection gained during our study can be used for further research.

Study design
The study was performed between August 2016 and March 2020 among children age 12 to 59 months not vaccinated with any pneumococcal vaccine and attending a 'non-sick-visit' in hospital outpatients' clinics or community healthcare centers in cities of Warsaw and Wroclaw. First, parents (or child's legal guardians) were asked if family would be interested in participation in the study. If they responded positively, they were informed about the study goals and procedures and asked to give written informed consent for the child's participation. Next, parents were asked to ll-in the questionnaire and provide information on the child's age, gender, environment (number of siblings, day-care attendance, presence of smoker in child's household), and clinical information (pneumococcal vaccination, reason of doctor's o ce visit, occurrence of chronic diseases, symptoms of lower respiratory tract infections, antibiotic therapy in the past three months). The questionnaire was reviewed on site by the study personnel to exclude children that have been vaccinated with any pneumococcal vaccine, were treated within last 4 weeks with any antibiotic, have any immunode ciency or symptom of lower respiratory tract infections. Finally, a saliva sample and nasopharyngeal swab were collected from each child by the study medical personnel. Sheep blood (Graso, Poland) with 5 µg/ml gentamycin (Sigma-Aldrich, USA) (GENT-agar) and incubated for 18-24h in 35°C, 5% CO 2 as previously described [38,20]. Once pneumococcus-like colonies were re-plated, all remaining colony growth was harvested [20]. These harvests represented samples culture-enriched (CE) for S. pneumoniae [38]. Replated isolates were identi ed as S. pneumoniae based on susceptibility to optochin (BioMerièux, France) and bile solubility (Becton Dickinson, USA) [39].   Correlations between results of pneumococcal serotypes detected by conventional culture and molecular method (qPCR). Panels (a) and (c) depict correlations between number of cultured (X-axis) and number of samples positive in qPCR among serotypes or serogroups targeted by qPCR assays. Panel (a) depicts results exclusively for NPS samples.
Panel (c) depicts results for all strains cultured from NPS or saliva versus detected in NPS or saliva in qPCR assays. Panel (b) shows correlation between NPS (X-axis) and saliva (Y-axis) for serotypes detected exclusively with qPCR assays. Serotypes not detected using a given approach have been assigned a value of 0.5. The rho, and p values have been calculated with Spearman's test.