Abstract
The associations of serum 25-hydroxyvitamin D (25(OH)D) with cirrhosis and liver cancer have not been well examined. It remained unclear whether functional vitamin D-related variants modify these associations. During follow-up of 401,551 UK Biobank participants with valid 25(OH)D data, we documented 4404 incident non-alcoholic fatty liver disease (NAFLD), 1989 incident cirrhosis, and 480 incident liver cancer. The binding affinity values of 25(OH)D to vitamin D binding protein (VDBP) were estimated based on combinations of VDBP variants rs4588 and rs7041. Vitamin D receptor (VDR) activity was calculated by summing the number of related alleles for rs11568820, rs2228570, and rs1544410. Hazard ratio (95% confidence intervals) for serum 25(OH)D of ≥50 nmol/L (vs. <25 nmol/L) were 0.70 (0.63-0.77) for NAFLD (ptrend = 1.37 × 10-14), 0.43 (0.38-0.49) for cirrhosis (ptrend = 8.66 × 10-29) and 0.65 (0.48-0.87) for liver cancer (ptrend = .010). We observed differential associations of 25(OH)D with hepatic fibrosis and cirrhosis (pinteraction = .024), and portal hypertension (pinteraction = .019) by VDR activity, with a stronger inverse association among participants with low VDR activity (ptrend = 1.28 × 10-9 for hepatic fibrosis and cirrhosis and ptrend = 1.52 × 10-14 for portal hypertension). The association of 25(OH)D with liver cancer differed by GC isoforms (pinteraction = .033), with a stronger inverse association among participants with low affinity isoforms (ptrend = .001). Serum 25(OH)D appears to prevent NAFLD, cirrhosis, and liver cancer, especially for participants with low VDR activity/low affinity GC isoforms.