Mixed-Effects Physiologically Based Pharmacokinetic Modeling of Methylmercury across the Maternal–Infant Continuum: Multi-Matrix Evaluation in Two Korean Birth Cohorts

Abstract

Prenatal methylmercury (MeHg) exposure drives neurodevelopmental impairment in East Asia, but existing physiologically based pharmacokinetic (PBPK) models rely on population-average parameters and single-matrix validation. We developed a three-submodel mixed-effects PBPK (MEM-PBPK) framework validated against maternal blood, cord blood, urinary, and breast milk mercury in two Korean birth cohorts (Ko-CHENS, n = 3,625; NOEMOC, n = 242). Clinical chemistry-individualized Michaelis–Menten placental kinetics reproduced second-trimester blood mercury (mean fold error, MFE = 0.946); empirical Bayes random effects improved cord blood prediction to MFE = 0.954 (within-2-fold = 100%). Biomarker bridging replicated +38.7% raw and −32.6% creatinine-corrected urinary trimester changes (p < 0.001), and breast milk MFE = 1.057. Critically, postnatal glomerular filtration maturation prolonged neonatal inorganic mercury renal half-life ∼5-fold (∼160 d at birth vs ∼32 d in adults; partial rank correlation coefficient, PRCC > 0.99) without affecting blood MeHg, demonstrating that blood-only biomonitoring systematically underestimates neonatal kidney mercury burden. These findings support cord-blood-anchored multimatrix biomonitoring and postnatal urinary iHg surveillance in East Asian populations.