BACKGROUND: Coronary flow reserve (CFR) by adenosine echocardiography in left anterior descending (LAD) or posterior descending coronary arteries may predict clinical outcome. METHODS: We used models accounting (Cox's model) or not (logistic regression and neural network) for time to event and either considered (forced models) or not (stepwise logistic regression and neural network models) all among 21 covariates to predict 1-year composite events after LAD CFR. RESULTS: There were 553 consecutive patients with coronary artery disease (CAD): 89 patients had also posterior descending CFR. During 1-year follow-up 328 patients were event-free, 35 had composite ischemic events and 190 underwent short-term revascularization. LAD and posterior descending CFR (respectively, 1.53 ± 0.83, N = 225 and 1.84 ± 0.80, N = 42) were significantly (P < 0.0001) lower in patients with events (or with revascularization following CFR measurement) than in those without (respectively, 3.13 ± 0.84, N = 328, and 2.53 ± 0.72, N = 47). Using LAD CFR as a continuous covariate, by both forced Cox's and logistic regression, coefficients (t values, respectively, -14.11 and -10.19) were significant (both P < 0.00001) to predict outcome. Global predictive accuracies by neural network, adopting a receiver operating characteristic areas under the curve (ROC) assessment, were excellent (>0.91) and the role of LAD CFR among predictors was overwhelming. Other indices of myocardial ischemia and the presence of coronary stenoses or previous infarction did not modify the multivariable predictive role of LAD CFR. When patients with revascularization were discounted, the LAD CFR predictive role was the same. CONCLUSIONS: Thus, adenosine echocardiography-based LAD CFR predicts 1-year composite ischemic events in patients with CAD, independent of the multivariable model adopted. Posterior descending CFR also has a role.
BACKGROUND: Coronary flow reserve (CFR) by adenosine echocardiography in left anterior descending (LAD) or posterior descending coronary arteries may predict clinical outcome. METHODS: We used models accounting (Cox's model) or not (logistic regression and neural network) for time to event and either considered (forced models) or not (stepwise logistic regression and neural network models) all among 21 covariates to predict 1-year composite events after LAD CFR. RESULTS: There were 553 consecutive patients with coronary artery disease (CAD): 89 patients had also posterior descending CFR. During 1-year follow-up 328 patients were event-free, 35 had composite ischemic events and 190 underwent short-term revascularization. LAD and posterior descending CFR (respectively, 1.53 ± 0.83, N = 225 and 1.84 ± 0.80, N = 42) were significantly (P < 0.0001) lower in patients with events (or with revascularization following CFR measurement) than in those without (respectively, 3.13 ± 0.84, N = 328, and 2.53 ± 0.72, N = 47). Using LAD CFR as a continuous covariate, by both forced Cox's and logistic regression, coefficients (t values, respectively, -14.11 and -10.19) were significant (both P < 0.00001) to predict outcome. Global predictive accuracies by neural network, adopting a receiver operating characteristic areas under the curve (ROC) assessment, were excellent (>0.91) and the role of LAD CFR among predictors was overwhelming. Other indices of myocardial ischemia and the presence of coronary stenoses or previous infarction did not modify the multivariable predictive role of LAD CFR. When patients with revascularization were discounted, the LAD CFR predictive role was the same. CONCLUSIONS: Thus, adenosine echocardiography-based LAD CFR predicts 1-year composite ischemic events in patients with CAD, independent of the multivariable model adopted. Posterior descending CFR also has a role.
Authors: Li-Ming Gan; Sara Svedlund; Ann Wittfeldt; Charlotte Eklund; Sinsia Gao; Göran Matejka; Anders Jeppsson; Per Albertsson; Elmir Omerovic; Amir Lerman Journal: J Am Heart Assoc Date: 2017-04-18 Impact factor: 5.501