Myocardial perfusion imaging can predict outcomes in cardiac patients. However, limited data exist regarding its prediction of cardiovascular outcomes in cancer patients. We sought to determine whether myocardial perfusion imaging predicts long-term cardiovascular outcomes in cancer patients.We performed a retrospective review of 787 consecutive patients at our institution who underwent myocardial perfusion imaging from January 2001 through March 2003. The Cox proportional hazard model was applied, and total cardiac events, cardiac death, and all-cause death were determined for 3 years. We considered P <0.05 to be statistically significant.Patients with abnormal myocardial perfusion imaging results were more likely to be male and older, with heart disease, more vascular risk factors, and lower left ventricular ejection fraction (0.52 +/- 0.14 vs 0.63 +/- 0.11; P <0.001) than patients with normal myocardial perfusion imaging results. Multivariate predictors of total cardiac events included age (P = 0.023), hyperlipidemia (P = 0.0021), pharmacologic myocardial perfusion imaging (P <0.01), left ventricular ejection fraction (P <0.001), and abnormal myocardial perfusion imaging (P = 0.012). Multivariate predictors of cardiac death included age (P = 0.026) and left ventricular ejection fraction (P = 0.0001). Multivariate predictors of all-cause death were age (P = 0.0001), atrial fibrillation (P = 0.0012), and smoking (P <0.001). Overall survival was improved when patients took aspirin (P = 0.0002) and upon each unit increase in left ventricular ejection fraction (P <0.001).Myocardial perfusion imaging in cancer patients can predict 3-year cardiac outcomes. Increasing age, atrial fibrillation, and smoking were associated with worse outcomes, whereas higher left ventricular ejection fraction and the taking of aspirin were protective.
Keywords: Aspirin/therapeutic use; comorbidity; coronary disease/epidemiology/radionuclide imaging; death, sudden, cardiac/prevention & control; electrocardiography; exercise test/methods/statistics & numerical data; heart/radionuclide imaging; multivariate analysis; neoplasms/epidemiology/mortality; predictive value of tests; proportional hazards model; retrospective studies; risk assessment/methods; ventricular function, left.