Ability of the prognostic model of J-ACCESS study to predict cardiac events in a clinical setting: The APPROACH study

Abstract Background In patients with coronary artery disease (CAD), one of the risk models available in Japan was a multivariate risk prediction model based on a Japanese multicenter database: the Japanese Assessment of Cardiac Events and Survival Study by Quantitative Gated SPECT (J-ACCESS). The aim of this study was to clinically validate the accuracy of this risk model. Methods We evaluated the performance of the J-ACCESS model using data derived from the Assessment of the Predicted value of PROgnosis of cArdiaC events in Hokuriku (APPROACH) registry. Variables of age, summed stress score (SSS), left ventricular ejection fraction (LVEF), estimated glomerular filtration rate (eGFR), and diabetes mellitus were included. The major cardiac events were defined as cardiac death, non-fatal myocardial infarction, and heart failure that required hospitalization. The patients were followed up for three years to compare between predicted risk and actual events. Results We evaluated 283 patients with suspected or confirmed CAD receiving myocardial perfusion imaging using 99m Tc-tetrofosmin between March 2009 and August 2011. Mean age was 68.9 ± 10.1 years, mean eGFR 67.4 ± 24.3 mL/min/1.73 m 2 , mean SSS 5.2 ± 7.2, and mean LVEF 65.4 ± 14.0%. Fourteen (4.9%) patients experienced major cardiac events including cardiac death in 4 patients (1.4%), non-fatal myocardial infarction in 1 patient (0.3%), and severe heart failure in 9 patients (3.2%), respectively. While SSS ≥ 8, LVEF  2 , and event risk ≥ 10% were significant variables in survival analysis, multivariate proportional hazard analysis showed that only LVEF and eGFR were significant. The event rate estimated from the J-ACCESS model was comparable to the actual number of major cardiac events (9 and 6, respectively, p  = 0.58 by Chi-square test). Conclusions The predictive ability of the J-ACCESS risk model is clinically valid among patients with CAD and could be applicable in clinical practice.