Abstract
Limited evidence exists on domestic water hardness (DWH) and incident type 2 diabetes (T2D), including the potential mediating role of metabolomic biomarkers. We aimed to assess the association between DWH and T2D, identify metabolomic biomarkers, and evaluate mediation by these biomarkers.We included 328,920 UK Biobank participants free of T2D and cancer at baseline. DWH exposure, including calcium (Ca), magnesium (Mg), and calcium carbonate (CaCO3), was derived from water-utility outputs at baseline. Incident T2D was ascertained from linked health records. Baseline nuclear magnetic resonance profiling quantified 168 plasma biomarkers. We used Cox regression to estimate associations between DWH and T2D risk, the elastic net model and the generalized boosts model to select and rank biomarkers, and causal mediation analyses to assess mediation by the selected biomarkers. Over a median follow-up of 13.6 years, 13,103 participants developed T2D. Higher Mg was inversely associated with incident T2D (Hazard ratio 0.95, 95% confidence interval 0.92-0.99), whereas Ca (1.01, 0.93-1.10) and CaCO3 (1.00, 0.93-1.06) showed no significant associations with T2D. In mediation analysis, cholesteryl esters in large HDL, triglycerides in LDL, tyrosine, GlycA, and glucose significantly mediated the Mg-T2D association, with proportions mediated ranging from 6.5% to 8.2%, and these biomarkers jointly mediated 17.2% of the association. These findings may deepen understanding of the mechanisms underlying the DWH-T2D association and inform prevention strategies.</p>