RESUMEN
OBJECTIVES: This study sought to determine whether arterial inflammation measured by (18)F-fluorodeoxyglucose positron emission tomography ((18)F-FDG-PET) improves prediction of cardiovascular disease (CVD) beyond traditional risk factors. BACKGROUND: It is unknown whether arterial (18)F-FDG uptake measured with routine PET imaging provides incremental value for predicting CVD events beyond Framingham risk score (FRS). METHODS: We consecutively identified 513 individuals from 6,088 patients who underwent (18)F-FDG-PET and computed tomography (CT) imaging at Massachusetts General Hospital between 2005 and 2008 and who met additional inclusion criteria: ≥30 years of age, no prior CVD, and free of cancer. CVD events were independently adjudicated, while blinded to clinical data, using medical records to determine incident stroke, transient ischemic attack, acute coronary syndrome, revascularization, new-onset angina, peripheral arterial disease, heart failure, or CVD death. FDG uptake was measured in the ascending aorta (as target-to-background-ratio [TBR]), while blinded to clinical data. RESULTS: During follow-up (median 4.2 years), 44 participants developed CVD (2 per 100 person-years at risk). TBR strongly predicted subsequent CVD independent of traditional risk factors (hazard ratio: 4.71; 95% confidence interval [CI]: 1.98 to 11.2; p < 0.001) and (hazard ratio: 4.13; 95% CI: 1.59 to 10.76; p = 0.004) after further adjustment for coronary calcium score. Addition of arterial PET measurement to FRS scores improved the C-statistic (mean ± standard error 0.62 ± 0.03 vs. 0.66 ± 0.03). Further, incorporation of TBR into a model with FRS variables resulted in an integrated discrimination of 5% (95% CI: 0.36 to 9.87). Net reclassification improvements were 27.48% (95% CI: 16.27 to 39.92) and 22.3% (95% CI: 11.54 to 35.42) for the 10% and 6% intermediate-risk cut points, respectively. Moreover, TBR was inversely associated with the timing of CVD (beta -0.096; p < 0.0001). CONCLUSIONS: Arterial FDG uptake, measured from routinely obtained PET/CT images, substantially improved incident CVD prediction beyond FRS among individuals undergoing cancer surveillance and provided information on the potential timing of such events.