Abstract
BACKGROUND: Observational studies have linked vitamin K and cancer, but the causality of this association remains unknown. This Mendelian randomization (MR) study aims to investigate the association between circulating phylloquinone (vitamin K(1)) levels and four female-specific cancers. METHODS: We used four single-nucleotide polymorphisms (SNPs) to instrument phylloquinone, with the reported F-statistic 16.00-28.44 for all variants. SNP-outcome associations were obtained from consortia meta-analyses, UK Biobank, and the FinnGen database (up to 145,257/419,675, 27,446/362,324, 15,181/591,477, and 2211/320,454 cases/controls for breast, ovarian, endometrial, and cervical cancer, respectively). Analyses were conducted using five complementary MR methods including pleiotropy robust approaches. The MR Egger intercept test, MR PRESSO global test and leave-one-out analyses were used to test for and identify pleiotropic variants. RESULTS: The relevance of the instrument was validated by positive control analyses on coagulation factor IX (p = 0.01). However, the main MR analysis and all sensitivity analyses were consistently supportive of a null association between phylloquinone and all four cancers (p > 0.05 for all analyses, across all methods). MR-PRESSO did not detect outlying variants, and there was no evidence of horizontal pleiotropy relating to any cancer outcome (p(intercept) > 0.26 for all). CONCLUSIONS: We found no evidence for an association between genetically predicted circulating phylloquinone levels and the risk of four female-specific cancers.