Very early life microbiome and metabolome correlates with primary vaccination variability in children

Abstract

We show that simultaneous study of stool and nasopharyngeal microbiome reveals divergent timing and patterns of maturation, suggesting that local mucosal factors may influence microbiome composition in the gut and respiratory system. Antibiotic exposure in early life as occurs commonly, may have an adverse effect on vaccine responsiveness. Abundance of gut and/or nasopharyngeal bacteria with the machinery to produce lipopolysaccharide-a toll-like receptor 4 agonist-may positively affect future vaccine protection, potentially by acting as a natural adjuvant. The increased levels of serum phenylpyruvic acid in infants with lower vaccine-induced antibody levels suggest an increased abundance of hydrogen peroxide, leading to more oxidative stress in low vaccine-responding infants.

Keywords: antibiotics; children; metabolome; microbiome; vaccine response.

Fig 1

Longitudinal samples collected from children in the first 2 years of life to understand the relationships between antibiotic usage, microbiome, and vaccine response. (a) Study design. Subjects were recruited within 3 weeks of birth and followed through to 2 years. Blood samples were taken where possible at 2 months, 1 year, and 2 years to measure antibody response against vaccination antigens. Subjects were scheduled to attend well-check clinic visits regularly as shown at the bottom of the panel. Samples were collected as possible at these visits as well as any sick visits. (b) Longitudinal sampling from subjects. For each subject in the study, a timeline showing their time in the study by age in days (gray horizontal lines), stool and nasopharyngeal samples collected (red and blue circles), vaccinations received (black vertical lines), oral antibiotics taken (green horizontal bars), and blood samples taken to measure antibody responses (black antibody symbols). (c) Normalized antibody titers at 1 year. A normalized antibody titer was calculated for all 19 antigens measured in the study from 72 subjects at 1 year of age. Antigens measured were those included in the DTaP/Hib vaccine [diphtheria (DT), tetanus (TT), pertussis toxoid (PT), filamentous hemagglutinin (FHA), pertactin (PRN), and Haemophilus influenzae type b polysaccharide (PRP)] and 13 serotypes of Streptococcus pneumoniae included in the Prevnar 13 pneumococcal conjugate vaccine (PCV) (serotypes 1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, and 23F). Median normalized titers were also calculated across vaccines, for the DTap/Hib vaccine only, and for the PCV only, for normal vaccine responders (NVR) indicted in purple, and low vaccine responders (LVR) indicated in green.

Fig 2

Longitudinal sampling allows for investigation of microbiome development with age. (a) Stool microbiome development. The Kraken genus alpha (left) and beta (right) diversity colored by the age of the subject when the sample was collected. (b) Nasal microbiome development. The nasal operational taxonomic unit (OTU) alpha (left) and beta (right) diversity colored by the age of the subject when the sample was collected. (c) Dense sampling from a study subject. Taxa bar plots of nasal (relative OTU abundance grouped by taxonomic order, top) and stool (relative Kraken order abundance, bottom) microbiome samples from subject 108 by age in days. Circles at the bottom indicate whether oral antibiotics were taken within 7 days of the sample (top row, black) and the type of clinic visit (pink-red circles, bottom).

Fig 3

Early predictors of vaccine response identified in the 2-month microbiome and metabolome. (a) Enrichment analysis showing KO modules at 2 months associated with cross titer vaccine response at 1 year. KO module enrichment analysis was done to find modules significantly enriched (adjusted P-value <0.05) with KOs nominally significantly correlated with cross-vaccine titer (P < 0.05). Top hits are shown with x-axis showing the ratio of genes in the enriched module to the total number of genes in the analysis. (b) Correlations between metabolite abundances at 2 months and vaccine response at 1 year.