Abstract
Background: Following vitamin D(3) oral administration, attained serum concentrations of its metabolite 25-hydroxyvitamin D(3) (25(OH)D(3)) are variable among children. Methods: We developed physiologically based pharmacokinetic (PBPK) modelling using annually measured serum 25(OH)D(3) concentrations in 77 Cape Town schoolchildren aged 6-11 years who received weekly oral doses of 10,000 IU vitamin D(3) for 3 years during a clinical trial (Δ25(OH)D = 32.2 nmol/L, 95% CI: [-3.2, 65.8] nmol/L). Simulations were performed to test the model on 463 other participants in the same trial, and in a cohort of 1756 Mongolian schoolchildren aged 6-11 years who received weekly oral doses of 14,000 IU vitamin D(3) for 3 years in another trial. Results: The best model attributed most of the variability in post-supplementation 25(OH)D(3) concentrations to hepatic clearance and covariates including weight (ΔAIC = -21) and ZBMI (body mass index Z-score, ΔAIC = -34). For 463 other children from the Cape Town trial (Δ25(OH)D = 25.8 nmol/L, 95% CI: [8.3, 47.2] nmol/L), mean estimation error was 5.3 nmol/L, and 76.7% of observations were within the 95% prediction intervals. Our simulation supported the previous proposal that serum 25(OH)D(3) should exceed 50 nmol/L among 97.5% of European children at 24.4 μg/day vitamin D(3) dosing. At a higher weekly dose (14,000 IU), the Mongolian children demonstrated a higher average increase in serum 25(OH)D(3) (40.6 [-2.9, 88.9] nmol/L) but were overestimated by the model. Conclusion: We developed the first PBPK model to successfully predict the long-term serum 25(OH)D(3) increases in healthy schoolchildren in Cape Town who received orally administered vitamin D(3) and exhibited higher relative increases than Mongolian children.