Prevalence of Streptococcus pneumoniae in conjunctival flora and association with nasopharyngeal carriage among children in a Vietnamese community

Conjunctival pneumococcal serotypes among members of a community have not been investigated well. We determined the prevalence and association of Streptococcus pneumoniae in the nasopharynx and conjunctiva among children in a community before pneumococcal conjugate vaccine introduction. In October 2016, conjunctival and nasopharyngeal swabs were collected from children (< 24 months old) and nasopharyngeal swabs from mothers in Nha Trang, Vietnam. Quantitative lytA PCR and DNA microarray were performed to detect and serotype S. pneumoniae. The association between S. pneumoniae in the nasopharynx and conjunctiva was evaluated using multivariable logistic regression model. Among 698 children, 62 (8.9%, 95% CI 6.9–11.2%) were positive for S. pneumoniae in the conjunctiva. Non-encapsulated S. pneumoniae were most commonly identified, followed by serotypes 6A, 6B, and 14. Nasopharyngeal and conjunctival detection were positively associated (aOR 47.30, 95% CI 24.07–92.97). Low birth-weight, day-care attendance, and recent eye symptoms were independently associated with S. pneumoniae detection in the conjunctiva (aOR 11.14, 95% CI 3.76–32.98, aOR 2.19, 95% CI 1.45–3.31, and aOR 3.59, 95% CI 2.21–5.84, respectively). Serotypes and genotypes in the conjunctiva and nasopharynx matched in 87% of the children. Three mothers’ nasopharyngeal pneumococcal samples had matched serotype and genotype with their child’s in the conjunctiva and nasopharynx. S. pneumoniae presence in nasopharynx and conjunctiva were strongly associated. The high concordance of serotypes suggests nasopharyngeal carriage may be a source of transmission to the conjunctiva.

Prevalence and factors associated with S. pneumoniae in the conjunctiva. Sixty-two children (8.9%, 95% CI 6.9-11.2%) had S. pneumoniae in the conjunctiva. Social demographics, clinical characteristics, and S. pneumoniae carriage in the child's and mother's nasopharynx among children with and without conjunctival S. pneumoniae are shown in Table 1. Respiratory hospitalization history, cough, runny nose, and eye symptoms in the last 2 weeks, day-care attendance, generally in the company of other children under five, and S. pneumoniae carriage in the child's and the mother's nasopharynx were positively associated with S. pneumoniae positive conjunctiva by univariable analysis.
Forty-seven (87%) of the 54 children with conjunctival pneumococcal detection had at least one serotype that matched with S. pneumoniae serotype identified in their nasopharynx. Additionally, these 47 matched serotype samples had the same genotype by array CGH analysis of the genome component of the microarray. For the five mother's nasopharyngeal samples of children with conjunctival pneumococcal detection, three samples matched by serotype and genotype with the conjunctival and nasopharyngeal samples of their children (

Discussion
To our knowledge, this study is the first to report a positive association between conjunctival S. pneumoniae detection and nasopharyngeal S. pneumoniae carriage. We conducted a large-scale community-based survey using advanced laboratory techniques for the prevalence of S. pneumoniae in a community.
There have been only a few studies describing the prevalence of S. pneumoniae in conjunctivae, which were 2.2% among people aged from 0 to more than 65 years in China 6 , 2.7% among patients awaiting cataract surgery whose mean age was 71 years in Spain 25 , 3.2% among those aged 1 to 90 in the UK 7 , and 0% in the US (no age information) 8 . In this study, the prevalence of S. pneumoniae in the conjunctiva among unvaccinated children aged < 24 months in a community was 8.9% (95% CI 6.9-11.2%). The higher prevalence of S. pneumoniae in our study may be due to several reasons including participants' age, geography, and season or climate, considering previous reports as follows. Tao et al. 6 reported a prevalence of 4.2% in children 0-6 years, and of 1.6% among those from 7 to more than 65 years old. Their results were consistent with an earlier study that found a significant difference (p < 0.002) in the prevalence of Streptococcus species in the conjunctivae of children (aged 17 years Table 1. Effect of each characteristic on having S. pneumoniae in conjunctiva, estimated using logistic regression models. CI confidence interval, SD standard deviation. a Odds ratio of conjunctival S. pneumoniae detection for nasopharyngeal S. pneumoniae carriage adjusted by sex, age group (< 12 months or 12-23 months), cough, runny nose, and eye symptom (eye discharge, red eye, or itching) history in the last 2 weeks, low birth-weight, child's day-care attendance, number of family members in the household, presence of smokers in the household, and maternal pneumococcal carriage, considering clustering in each commune.   NT2-CGH-01 14-CGH-01;   www.nature.com/scientificreports/ or less) and adults (14.9% versus 2.2%) 9 . The differences in S. pneumoniae prevalence by age, especially between adults and children, may be attributed to several potential mechanisms; including age-related changes in general immune responsiveness, tear composition and dynamics, patterns of exposure to bacteria, past antibiotic utilization, and the flora of adjacent areas, such as the skin and upper respiratory tract 9 . The difference in the prevalence of microorganisms in conjunctival flora in relation to geography was described as early as 1954 when results from eye cultures from different London based eye hospitals varied 7,26 . Seasonal variation in the same area may be another important factor. This is supported by a unique study of 4432 patients undergoing cataract surgery between 1994 and 1996 in Madrid, which showed rising isolation rates of S. pneumoniae in March, November, and December in ocular surface flora 25 . Also, these previous studies were conducted in temperate region and the bacteria prevalence may be different in tropical as this study. Finally, the methods for detecting S. pneumoniae in the conjunctiva of subjects from a community differed between the studies; while the previous studies used mostly culture techniques [6][7][8][9] , this study used highly sensitive and specific approaches, including lytA quantitative PCR (qPCR), culture, and DNA microarray 27 . Previously, several studies evaluated correlations among isolates from conjunctivae, middle ear fluids, and nasopharynges in conjunctivitis-otitis syndrome [28][29][30] . Bingen et al. used restriction fragment length polymorphism to show that paired conjunctiva and middle ear fluid isolates of Haemophilus influenzae were identical 29 . Groothuis et al. explored the correlation between conjunctival and nasopharyngeal cultures among conjunctivitis-otitis media syndrome 30 . However, there have been no studies demonstrating the association of S. pneumoniae in conjunctivae and nasopharynges in members of a community. Here, we report the association between S. pneumoniae in the conjunctiva and the nasopharynx.
S. pneumoniae is a causative agent of many documented conjunctivitis outbreaks, and NT pneumococci are commonly identified as the etiological agent [31][32][33] . The association of NT S. pneumoniae with conjunctivitis was first suggested by Finland and Barnes in 1977 32 . Indeed, NT S. pneumoniae have caused several conjunctivitis outbreaks in the USA 34,35 , and were also recognized as a frequent cause of sporadic cases of conjunctivitis in Spain 22 . Previous studies found that 23-90% of S. pneumoniae isolates causing conjunctivitis were NT, with higher rates after PCV programs were introduced 22,31,33 . In studies using phenotypic testing, NTs may include both nonencapsulated S. pneumoniae as well as isolates for which no serotype was determined (e.g. due to the isolate not expressing capsule, or the test not including all serotypes). Keller et al. emphasized that surface proteins unique to non-encapsulated S. pneumoniae enhanced colonization and virulence despite the lack of a capsule 33 . Our study determined that 59% of S. pneumoniae-positive conjunctival samples were NeSp, the majority of which were NT2 that includes the NspA/PspK surface protein encoded by the null capsule locus 24 . Besides, 37% of PCV13-type S. pneumoniae might imply the effect of PCV introduction on reducing the prevalence in conjunctiva and also incidence of pneumococcal conjunctivitis in this area.
Although S. pneumoniae in nasopharynx and conjunctiva showed concordance in serotype and genotype among matched samples from children, the serotype distribution in the two sites was different. The most frequent conjunctival serotypes in this study were NeSp, 6A, 6B, 14, 23F, and 19F, among which NeSp had a markedly high (59%) and 6A had low (12%) proportion compared to those in the nasopharynx (22% and 29%, respectively) among the same subjects. Specific conjunctival factors, including tissue immune responsiveness, tear composition (immunoglobulins and antimicrobial enzyme lysozyme), dynamics (mechanical action of the eyelids), and patterns of exposure to bacteria might have an important role in serotype distribution differences between the conjunctiva and the nasopharynx 9,25 .  www.nature.com/scientificreports/ In our study, the lower rate of conjunctival S. pneumoniae detection than nasopharyngeal carriage, and the concordance of serotype and genotype in matched samples, might indicate that colonization of S. pneumoniae starts in the nasopharynx and spreads to the conjunctiva. Also, the tear composition and dynamics explained above might make it more difficult for potentially pathogenic microorganisms to colonize the conjunctival surface 9 . S. pneumoniae can be plausibly transmitted from the nasopharynx to the conjunctiva through the nasolacrimal duct, by the retrograde passage of fluid from the nose to the conjunctiva during nasal congestion and because of short ducts during infancy and early childhood. Also, S. pneumoniae may reach the conjunctival sac from the skin, the surrounding environment, hand contamination, and through contact with persons such as the mother 36 . The low nasopharyngeal carriage prevalence of mothers in this study indicates that they were unlikely to be the major source of transmission. We did not investigate the conjunctival S. pneumoniae prevalence in mothers in this study; however, it would be interesting to consider the same in a future study with a larger sample size to determine the prevalence in mothers and the potential transmission route.
To our knowledge, this is the first study to investigate risk factors related to conjunctival S. pneumoniae detection in children. Whilst we anticipated that age would be a risk factor for conjunctival S. pneumoniae detection, we did not find a difference in our study, most likely because the paediatric groups were relatively similar in age. Day-care attendance increased conjunctival S. pneumoniae detection in this study. Day-care attendance is a wellestablished factor associated with nasopharyngeal S. pneumoniae carriage 16,19,37,38 . Our results are in accordance with Sthapit et al., who showed that conjunctival bacterial flora is similar in either sex 39 . This study also determined that low birth-weight increased S. pneumoniae detection in the conjunctiva in children < 24 months. This may be because of the susceptibility of low birth-weight children to infectious diseases, compared to normal birth weight children, due to poor immune responses 40 . In contrast, the decreased nasopharyngeal carriage risk with low birth-weight was a surprising finding, though this result has been reported in Dutch and Indian babies 19,41 .
There are some limitations to the study. First, this was a cross-sectional study conducted in October 2016, and results may vary by season. Longitudinal studies may also provide insight as to whether detection of S. pneumoniae in the conjunctiva of children in a community is a true carriage, such as observed in the nasopharynx. In a few cases, the prevalence of pneumococcal strains was determined by positive lytA detection only. It should be considered that other related streptococcal species may also have lytA homologues and could yield false-positive results. Nevertheless, this accounted for a minor proportion, which would not have affected the findings of this study. In addition, we did not assess the presence of viruses or other bacterial species in this study.
Nearly 9% of children in a central Vietnamese community aged < 24 months had S. pneumoniae in the conjunctiva before introduction of PCV in the community. We have reported the novel finding that S. pneumoniae in the conjunctiva was positively associated with that in the nasopharynx, and that the serotypes and genotypes in the conjunctiva mostly matched those in the nasopharynx for each child. These findings suggest that the nasopharynx may be a source of transmission for S. pneumoniae to the conjunctiva. Nevertheless, there was some evidence also for a different serotype distribution in nasopharynx and in conjunctiva of study subjects. In addition, there was a high proportion of NeSp in the conjunctiva. Our data provide further evidence that NeSp has a unique ability to colonize in the ocular environment.

Methods
Study site and participants. The study was conducted in six communes in the city of Nha Trang, central Vietnam. Nha Trang has 27 communes and each commune has one commune health center, providing a range of basic health services. In the study area, two communes were zoned as rural and four as urban areas, by the administrative boundary. A PCV-reduced dosing schedule trial has been initiated in this study site 42 , and the current study was a part of the pre-PCV baseline assessment of the trial (https ://clini caltr ials.gov/ct2/show/ NCT02 96123 1).
In each commune, 60 children, each from two target age groups, aged 4-11 months (younger group) and 14-23 months (older group), were randomly selected using the commune population registration records for enrolment into the study, regardless of current respiratory or eye symptoms. The populations of target age groups in this study, within the study area of the 6 communes, were 1004 for 4-11 months and 1590 for 14-23 months. The randomly selected children and their mothers were invited for examination and interview at the commune health center in each commune, in October 2016, before the introduction of PCV. Written informed consent was obtained from all parents or guardians before collecting samples and conducting interviews.
Data, sample collection, and testing. A trained healthcare worker in each commune health center interviewed each participant's mother to collect demographic, socioeconomic, and clinical information using a structured interview form. Data collected included: sex, date of birth, birth-weight, gestational age at birth, if the child has ever been hospitalized for respiratory disease, PCV history, current symptoms in the last 2 weeks (cough, runny nose, eye symptoms including eye discharge, red eye, itchy eye, and others), breast feeding history, child's day-care attendance, whether the child is usually living with other children aged under five, education level of the mother, household income last month, family members smoking, household having farm animals, and residential area. Births were defined as preterm if children were born < 37 gestational weeks, and of low birth-weight if the birth-weight was < 2500 g, following the guidelines in the International Classification of Diseases-10: version 2010 43 . The child's conjunctival and nasopharyngeal swabs, and the mother's nasopharyngeal swabs, were obtained by a doctor using a nylon flocked swab and stored in skim milk, tryptone, glucose, and glycerol (STGG) medium according to WHO guidelines 44 .
Swabs were sent to the Pasteur Institute Nha Trang, where DNA extraction and real-time qPCR targeting lytA of S. pneumoniae was performed. Samples that were positive (Ct value < 35) or equivocal (Ct value [35][36][37][38][39][40] by qPCR-testing were cultured on selective agar, and DNA was extracted from bacterial growth on the QIAcube Sample size. Sample size was calculated based on the primary outcome of presence of conjunctival S. pneumoniae. We assumed the true proportion of children aged < 24 months in the study area who have S. pneumoniae in the conjunctiva to be 0.5 to obtain the largest sample size, and required the estimate to be within 0.04 of the true value (precision) within the 95% confidence interval (CI). Then, the minimum sample size was calculated to be 600.

Ethical considerations.
Institutional Review Boards at the National Institute of Hygiene and Epidemiology, Hanoi, Vietnam (the approval number VN01057), and the Institute of Tropical Medicine, Nagasaki University, Nagasaki, Japan (151203149-2), approved this study. All methods were performed in accordance with the relevant guidelines and regulations.
Statistical analysis. We calculated the prevalence of conjunctival and nasopharyngeal S. pneumoniae among children < 24 months and that of nasopharyngeal S. pneumoniae among their mothers in the study area. S. pneumoniae serotype distribution in the conjunctiva and nasopharynx were shown graphically and the concordance was evaluated. We showed the number and the proportion of children with the same S. pneumoniae serotype/genotype in the nasopharynx and in the conjunctiva. We compared demographic, socioeconomic, and clinical characteristics, and S. pneumoniae prevalence in the nasopharynges of the children and mothers between children with and without conjunctival S. pneumoniae. Crude odds ratios of conjunctival S. pneumoniae detection were analyzed for each characteristic using logistic regression. Based on previous studies of risk factors associated with conjunctival bacterial prevalence 6,39 or nasopharyngeal S. pneumoniae carriage 16,19,46,47 , the following potential confounders were selected a priori and included in a model to estimate the effect of the child's nasopharyngeal S. pneumoniae carriage on conjunctival S. pneumoniae detection: sex, age group (< 12 months or 12-23 months), cough, runny nose, and eye symptoms (eye discharge, red eye, or itching) history in the last 2 weeks, low birth-weight, day-care attendance, number of family members in the household, presence of smokers in the household, and maternal pneumococcal carriage. CIs were adjusted for the clustering of communes using robust standard errors. Statistical analyses were conducted using Stata version 14.0 software (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP).

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/