Previously, we demonstrated that prebiotics may provide a complementary strategy for increasing calcium (Ca) absorption in adolescents which may improve long-term bone health. However, not all children responded to prebiotic intervention. We determine if certain baseline characteristics of gut microbiome composition predict prebiotic responsiveness. In this secondary analysis, we compared differences in relative microbiota taxa abundance between responders (greater than or equal to 3% increase in Ca absorption) and non-responders (less than 3% increase). Dual stable isotope methodologies were used to assess fractional Ca absorption at the end of crossover treatments with placebo, 10, and 20 g/day of soluble corn fiber (SCF). Microbial DNA was obtained from stool samples collected before and after each intervention. Sequencing of the 16S rRNA gene was used to taxonomically characterize the gut microbiome. Machine learning techniques were used to build a predictive model for identifying responders based on baseline relative taxa abundances. Model output was used to infer which features contributed most to prediction accuracy. We identified 19 microbial features out of the 221 observed that predicted responsiveness with 96.0% average accuracy. The results suggest a simplified prescreening can be performed to determine if a subject's bone health may benefit from a prebiotic. Additionally, the findings provide insight and prompt further investigation into the metabolic and genetic underpinnings affecting calcium absorption during pubertal bone development.
Keywords: Adolescence; Bone health; Calcium absorption; Machine learning; Mineral metabolism; Prebiotic microbiome; Support vector machine.
© 2024. The Author(s).