Nonlinear Association of Total Bilirubin with Left Ventricular Mass in Untreated Essential Hypertension: A Secondary Cross-Sectional Analysis

Background: Higher serum bilirubin has been associated with lower left ventricular mass in hypertension, but whether the relation is nonlinear and whether reported thresholds are reproducible remain unresolved. Methods: We performed a secondary cross-sectional analysis of a publicly archived cohort of 339 untreated hypertensive patients; 327 had complete data. Left ventricular mass index (LVMI) was the primary outcome and left ventricular hypertrophy (LVH) the secondary outcome. We used multivariable regression and restricted cubic splines, and assessed change-point precision by profile likelihood and bootstrap resampling. Results: Each 1 μmol/L higher total bilirubin was associated with a 1.20 g/m2 lower LVMI (95% CI, −1.73 to −0.67; p < 0.001), and the association was nonlinear (p < 0.001), with a steeper slope below the median (p for interaction <0.001). The change point was imprecisely estimated (maximum likelihood 12.5 μmol/L; profile-likelihood interval 10.4–13.1), and the association was confined to participants at or above the median fasting glucose (q = 0.004). Adding bilirubin increased the optimism-corrected C-statistic by 0.056. Conclusions: Thresholds should be interpreted as descriptions of curve shape rather than reproducible cut points, and the present evidence does not support using bilirubin for risk stratification.