Disproportionality Analysis of Breast Cancer-Related Reports Across Vitamin K Homologs in the FDA Adverse Event Reporting System

Background/Objective: Vitamin K (VK) comprises a family of quinone compounds with potential involvement in cell death-related pathways through their redox properties. However, consistent findings have not been obtained regarding the clinical significance of VK in breast cancer (BC). Thus, we used the FDA Adverse Event Reporting System (FAERS) to examine the co-reporting patterns of BC-related reporting terms among VK-related reports. Methods: FAERS reports from 2004 Q1 to 2024 Q3 were analyzed using narrow-scope Preferred Terms from two BC-related Standardized MedDRA Queries. Reporting odds ratios (RORs), proportional reporting ratios (PRRs), Fisher’s exact tests, and expected co-report counts under independence were calculated for VK overall and individual homologs, followed by exploratory comparisons with other quinone-containing compounds. An INDI-based sensitivity analysis excluded reports with BC-related indication terms. Results: Among 32,156 VK-related reports, 136 were BC-related. VK overall showed significantly lower reporting disproportionality (ROR, 0.486; 95% confidence interval [CI], 0.411–0.575). The predefined criteria for lower reporting disproportionality were also met for phytomenadione (VK1), menatetrenone (MK-4), the integrated VK2 category, and menadione (VK3). For VK3, 2 co-reports were observed compared with 13.59 expected (ROR, 0.183; 95% CI, 0.053–0.631). In the INDI-based sensitivity analysis, the lower reporting pattern remained for VK overall and VK1, was retained but borderline for the integrated VK2 category, and was not retained for MK-4 or VK3. Conclusions: In FAERS, BC-related terms were infrequently co-reported with VK-related reports (ROR < 1), with this pattern being partially retained in the INDI-based sensitivity analysis. This reporting pattern does not indicate reduced incidence, a protective or antitumor effect, or any causal relationship. However, it may reflect the indication structure of the comparator reporting population and other sources of reporting biases. These descriptive observations are presented solely to generate hypotheses for future mechanistic research and analytical epidemiological studies with more rigorous confounding control.