Background: Rotavirus gastroenteritis remains a leading cause of morbidity and mortality despite the introduction of vaccines. Research shows there are several factors contributing to the reduced efficacy of rotavirus vaccines in low- and middle-income settings. Proposed factors include environmental enteric dysfunction (EED), malnutrition, and immune dysfunction. This study aimed to assess the effect of these factors on vaccine responses using a machine learning lasso approach.

Methods: Serum samples from two rotavirus clinical trials (CVIA 066 n = 99 and CVIA 061 n = 124) were assessed for 11 analytes using the novel Micronutrient and EED Assessment Tool (MEEDAT) multiplex ELISA. Immune responses to oral rotavirus vaccines (Rotarix, Rotavac, and Rotavac 5D) as well as a parenteral rotavirus vaccine (trivalent P2-VP8) were also measured and machine learning using the lasso approach was then applied to investigate any associations between immune responses and environmental enteric dysfunction, systemic inflammation, and growth hormone resistance biomarkers.

Results: Both oral and parenteral rotavirus vaccine responses were negatively associated with retinol binding protein 4 (RBP4), albeit only weakly for oral vaccines. The parenteral vaccine responses were positively associated with thyroglobulin (Tg) and histidine-rich protein 2 (HRP2) for all three serotypes (P8, P6 and P4), whilst intestinal fatty acid binding protein (I-FABP) was negatively associated with P6 and P4, but not P8, and soluble transferrin receptor (sTfR) was positively associated with P6 only.

Conclusion: MEEDAT successfully measured biomarkers of growth, systemic inflammation, and EED in infants undergoing vaccination, with RBP4 being the only analyte associated with both oral and parenteral rotavirus vaccine responses. Tg and HRP2 were associated with responses to all three serotypes in the parenteral vaccine, while I-FABP and sTfR results indicated possible strain specific immune responses to parenteral immunization.