Abstract
Earlier studies suggest a protective association between vitamin K antagonist (VKA) anticoagulants and the incidence of cancer. The authors examined the associations between VKA therapy and incidence of 24 site-specific cancers with a Danish population-based cohort study, using heart valve replacement as an instrumental variable. The authors enrolled 9,727 Danish residents who received a replacement heart valve between 1989 and 2006. The heart valve recipients were matched with 95,481 unexposed individuals on age and sex. The authors used the heart valve replacement instrument to estimate rate ratios associating VKA therapy with incidence of the 24 site-specific cancers using Poisson regression models. Direct associations between VKA therapy and incidence of the 24 cancers were estimated in a prescription validation subset. The instrumental variable associations were plotted according to the inverse normal of rank percentile and subjected to semi-Bayes shrinkage adjustment for multiple comparisons. The pattern of associations was consistent with a null-centered Gaussian distribution. No individual cancer site showed a substantial positive or negative association with VKA therapy in the prescription validation subset, the instrumental variable analysis, or the analysis with semi-Bayes adjustment. These results do not support the existing hypothesis that VKA therapy is associated with reduced cancer risk.