Analysis of 5,731 Americans reveals that gamma-glutamyl transferase (GGT), a common liver enzyme, independently predicts 10-year hard cardiovascular events beyond traditional risk factors. The enzyme shows a linear dose-response relationship with heart attack and stroke risk, with effects partially mediated through elevated blood pressure (44.8%), blood sugar (19.0%), and altered cholesterol metabolism (13.4%). Machine learning models incorporating GGT alongside alkaline phosphatase and globulin achieved 0.751 area-under-curve accuracy in validation testing. This finding addresses a critical gap in cardiovascular risk assessment, where current prediction tools miss many individuals who later develop heart disease. GGT is routinely measured in standard liver panels, making implementation straightforward without additional testing costs. The liver-heart connection reflects shared metabolic pathways involving inflammation and oxidative stress that drive both hepatic dysfunction and arterial disease. However, this preprint awaits peer review, and results may change. The cross-sectional design cannot establish causation, and validation in diverse populations remains limited. If confirmed, incorporating readily-available liver biomarkers could substantially improve identification of high-risk individuals for preventive interventions.