Genetically predicted vitamin D levels and risk of head and neck cancer: a mendelian randomization study

Observational studies examining the association between vitamin D and head and neck cancer (HNC) have reported conflicting results. We conducted a two-sample MR study to investigate the causal association between genetically predicted serum 25-hydroxyvitamin D (25(OH)D) levels and the risk of HNC and its major subtypes: oral cavity, laryngeal, hypopharyngeal, and HPV-negative oropharyngeal cancers.

Genetic instruments were selected from GWAS of 441,291 individuals (UK Biobank). Cancer outcomes were obtained from the HEADSpAcE consortium. Multiple sclerosis (MS) was analyzed as a positive control. We identified 115 independent genetic instruments for serum 25(OH)D. Causal estimates were primarily derived using inverse variance weighted (IVW) MR, supported by MR-Egger, weighted median, and mode-based methods. Extensive sensitivity analyses assessed heterogeneity, pleiotropy, and directionality.

We identified 115 independent genetic instruments for serum 25(OH)D levels, which explained 5.12% of the total phenotypic variance. The mean F-statistic was 198.3 and satisfied MR assumptions. Only 17 of 682 (2.5%) SNP-confounder associations reached nominal significance (p  < 5 × 10⁻⁸), and none remained significant after Bonferroni correction, indicating no systematic confounding. Higher 25(OH)D levels were associated with a reduced risk of MS (IVW OR = 0.84, 95% CI 0.71-0.99, p = 0.038; weighted median OR = 0.82, 95% CI 0.70-0.96, p  = 0.015), while no evidence of a causal association was observed between 25(OH)D and risk of HNC or any subtype (all IVW p  > 0.05; ORs ranging from 0.95 to 1.25). Sensitivity analyses revealed no evidence of horizontal pleiotropy, influential outliers, or reverse causation for any cancer outcome.

This study provides robust genetic evidence that circulating 25(OH)D levels are not a major causal determinant of HNC risk and that our instruments are not confounded by major lifestyle or UV-related factors.