Menu
Menu
Menu
Menu
There is growing concern that interventions that alter microbial ecology can adversely affect health. We characterised the impact of the seven-valent pneumococcal conjugate vaccine (PCV7) on pneumococcal carriage and the bacterial component of the nasopharyngeal microbiome during infancy. Newborns were recruited into three groups as follows: Group1 (n = 33) was the control group and comprised infants who received PCV7 after 6 months and came from unvaccinated communities. Group 2 (n = 30) came from unvaccinated communities and Group 3 (n = 39) came from vaccinated communities. Both group 2 and 3 received PCV7 at 2, 3 and 4 months. Culture and 16 S rRNA gene sequencing were performed on nasopharyngeal specimens collected at regular intervals from infants. Nasopharyngeal carriage of PCV7 serotypes in Group 1 was significantly higher than in Group 2 and 3 (p < 0.01). However, pneumococcal carriage remained comparable due to an expansion of non-vaccine serotypes in Groups 2 and 3. Determination of phylogenetic dis(similarities) showed that the bacterial community structures were comparable across groups. A mixed effects model showed no difference in community richness (p = 0.15) and Shannon α-diversity (p = 0.48) across the groups. Immediate replacement of pneumococcal vaccine serotypes with non-vaccine serotypes may mitigate the impact of PCV7 on nasopharyngeal bacterial community structure and ecology.
Ethical approval to conduct this prospective longitudinal study was granted by the Gambian Government and Medical Research Council Unit The Gambia Joint Ethics Committee. Written, informed consent was obtained from the participants’ parents or guardian at recruitment. Enrolment of participants and all procedures were conducted in accordance with the relevant regulations and guidelines. The Western Region is representative of rural areas in The Gambia. HIV prevalence is estimated at 2% and the majority of villagers belong to the Jola, Mandinka and Fula ethnic groups29. Villages in the Western Region, The Gambia covering an area of approximately 90 km2 were selected for this study. Study participants were recruited from 27 villages each with estimated birth rates between three and twenty-six per year. The villages were split into 3 groups of 9 villages with estimated population sizes of 2000 and birth rates of approximately 80 per year. Trained village reporters in each village recorded and reported pregnancies, births, deaths and other serious events. Recruitment of subjects was carried out between November 2008 and April 2009 after written informed consent. To avert recruitment bias, all newborns from participating villages for whom informed parental consent was obtained were recruited into the study until each of the vaccination groups had at least 30 children. The study participants were recruited into three groups as outlined in the flowchart in Fig. 1. Group 1 and 2 participants were born in PCV7 unvaccinated villages while Group 3 participants were born in heavily vaccinated communities following the PCV7 vaccine trials conducted in 20062. Group 2 and 3 infants received three doses of PCV7 at 2, 3 and 4 months whereas Group 1 infants did not receive PCV7 at this age. PCV7 was introduced in the Gambia in August 2009, 11 months after the study was started. Group 1infants received at least 1 dose of PCV7 during the national catch-up campaigns. All Group 1 infants were at least 6 months old when they received the first dose of PCV. Group 2 and Group 3 infants also received additional doses of PCV7 following the implementation of PCV7 in The Gambia after 6 months of age. Nasopharyngeal swabs were collected within seven days of birth, then biweekly for the first six months and subsequently bi-monthly up to one year from all participants. Nasopharyngeal swabs were collected as previously described29 by trained field workers. The swab was immediately inoculated into a vial containing 1 mL of chilled skim milk–tryptone–glucose-glycerol (STGG) transport medium. The vials were kept on ice, transported to the MRCG Laboratories site in Fajara and stored at −70 °C within eight hours of collection. Pneumococci were isolated as previously described29. Nasopharyngeal swabs stored in STGG were thawed on ice and gently vortexed for five seconds. DNA was extracted from each swab using the PowerSoil® DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA, USA). 250 µL of the swab solution was transferred to the PowerBead® tubes and DNA was extracted following manufacture’s protocol. DNA was eluted in 100 µL of the kit elution buffer and immediately stored at −20 °C. DNA extractions were carried out in batches of 24 including one extraction control to which 250 µL of sterile DNAse free water was added instead. The V1-V3 region of the 16S rRNA gene were amplified using primers 5′-AGAGTTTGATCCTGGCTCAG-3′ and 5′-ATTACCGCGGCTGCTGG-3′ as previously described44, 45. The primers contained an adapter sequence and unique barcodes. The positive controls for all runs were purified. Anaerotruncus colhominis DNA, reagent controls and non-template controls were set up for each PCR run. Following amplification and purification, the amplicons were pooled at equimolar concentrations and sequenced on the 454 GS FLX Titanium Sequencing Platform (Roche, USA) as described elsewhere44, 45 at the Genome Institute (University of Washington in St. Louis, MO, USA). Quality control and data processing were performed using in-house pipelines at the Jackson Laboratory. Briefly, reads with length <200 bp and/or with more than a single ambiguous base call were discarded. Chimeric sequences and reads without the adapter sequences were also removed. Reads from the same samples were binned based on barcode and then the barcode, adapter and primer sequences at both terminals were trimmed. Samples with read depth less than 500 were discarded based on rarefaction analysis of all the samples. Alignment and taxonomic classification (Phylum to Genus) of the reads was carried out with the Ribosomal Database Project Naïve Bayesian Classifier using a 0.5 filter. At recruitment, background data including birth weight, birthplace, maternal age, maternal parity, ethnicity, sex, and vaccination status of mothers and siblings were collected. Socio-demographic and clinical information about the participants was collected at each visit including antibiotic use, anthropometric measurements, travel history, breastfeeding status, and infections (ear and chest). All data were collected on approved study forms and double entered and verified in an OpenClinica Electronic Data Capture system (Waltham, MA, USA) with an Access (Microsoft, Seattle, WA, USA) backend. Differences in baseline characteristics of the study participants recruited into the three study groups were tested using the Chi square test, Fisher’s exact test or Kruskal-Wallis test where appropriate. A mixed effects model was used to compare richness and Shannon α diversity across the study groups. Richness was positively skewed and was square root transformed for statistical analysis. Time-invariant risk factors (gender, place of birth, number of siblings, maternal age) and time-variant risk factors (age, vaccination period, season, breast feeding, travelling, antibiotic usage and nutritional status based on weight for height) were used as explanatory variables. Age and vaccination group were the main exposure risk factors and the remaining explanatory risk factors were considered as potential confounders. A random-intercept model was used to handle these two types of explanatory risk factors. All explanatory risk factors mentioned above were included in the fixed effects component of the model as well as pairwise and triple interaction terms between time, group and period. A backward model selection approach using the Wald test was adopted. Explanatory risk factors with a non-significant effect at the 5% significance level were discarded. Model residuals and random effects were checked for reliability. Kaplan Meier survival curves were used to compare the rates to first acquisition of pneumococcus and PCV7 serotypes in the three study groups. Statistical analyses were performed using STATA/SE 14.1, USA. Microbial community composition, structure and ecology analyses were performed using the Phyloseq package46 with R version 3.2.1. To reduce noise, low-occurrence, poorly represented OTUs were filtered out i.e. OTUs that did not appear more than once in 10% of the samples. To evaluate the effect of PCV7 vaccination group on overall microbial composition, the phylogenetic dis(similarities) among communities from the three groups were evaluated by metric multidimensional scaling (MDS)/ Principle component analysis (PCoA) on weighted UniFrac distance. Differential abundance of individual OTUs across the vaccination groups was explored using the DESeq package-extensions in the Phyloseq package. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
