Methods: A total of 100 asymptomatic patients with type 2 diabetes underwent electrocardiogram (ECG), echocardiography, biomarker assessment, and CMR at 3.0T including assessment of left ventricular ejection fraction and LGE. Global longitudinal strain from two- and four-chamber cines was measured using feature tracking.
Results: A total of 17/100 patients with no history of cardiovascular disease had silent MI defined by LGE in an infarct pattern on CMR. Only four patients with silent MI had Q waves on ECG. Patients with silent MI were older (65 vs 60, P = .05), had lower E/A ratio (0.75 vs 0.89, P = .004), lower GLS (–15.2% vs –17.7%, P = .004), and higher amino-terminal pro brain natriuretic peptide (106 ng/L vs 52 ng/L, P = .003). A combined risk score derived from these four factors had an area under the receiver operating characteristic curve of 0.823 (0.734–0.892), P < .0001. A score of more than 3/5 had 82% sensitivity and 72% specificity for silent MI.
Conclusions: Using measures that can be derived in an outpatient clinic setting, we have developed a novel screening tool for the detection of silent MI in type 2 diabetes. The screening tool had significantly superior diagnostic accuracy than current ECG criteria for the detection of silent MI in asymptomatic patients.