Abstract 9707: Risk Reclassification of Mortality by Exercise Myocardial Perfusion Imaging

Abstract

Background & Aim: There is growing interest in net reclassification methods for evaluating incremental value of myocardial perfusion imaging (MPI) over clinical markers in clinical risk stratification. In exercise testing, net reclassification improvement (NRI) analysis has not been used in large patient cohorts. We aim to assess prognostic and net reclassification value of exercise and MPI by single-photon emission computed tomography to estimate mortality. Methods: We analyzed a prospective cohort of 11,156 patients referred for MPI between 2006 and 2008 (mean age 59 SD 12 years) with no known coronary disease. Mean follow-up time was 2.5 years with all-cause mortality as the primary outcome. In the first analysis involving 11,156 patients, the effect of MPI result on mortality was determined using Cox regression models. We then applied reclassification methodologies to compare between a clinical model that included age, gender, coronary risk factors, ability to exercise, versus a clinical + MPI model that included summed stress score (SSS) and ejection fraction. 6556 (59%) of patients underwent treadmill exercise testing using the Bruce protocol with computation of the Duke Treadmill Score (DTS). In a second analysis that involved only this subgroup of patients who exercised, we compared the clinical model comprising of age, gender, risk factors, SSS, ejection fraction, with a model that included DTS. All risk-estimates were categorized as <1%, 1-3% and >3% risk of mortality in 4 years. Results: Mortality occurred in 491 patients(4.4%). In clinical risk-adjusted models, the ability to exercise significantly estimated all-cause mortality (HR=0.147, 95% CI= 0.114 to 0.191, p<0.001). Cox analysis showed that the addition of exercise improved incremental prognostic information (global chi-square increased from 639.58 to 758.02; p<0.001) and risk prediction (NRI =0.06; 95% CI=0.05 to 0.07, p<0.001). Amongst those who were able to exercise, the NRI by adding DTS was 0.08, 95%CI=0.04 to 0.12 (p<0.001). Conclusion: In this large cohort of patients, the ability to exercise and Duke Treadmill Score added modest but significant incremental prognostic and risk prediction value over routine clinical and MPI variables in estimating all-cause mortality.