A novel clinically relevant segmentation method and corresponding maximal ischemia score to risk-stratify patients undergoing myocardial perfusion scintigraphy

• Published in: Journal of nuclear cardiology 2014-08, Vol.21 (4), p.807-818
• Main Authors: Nudi, Francesco, Pinto, Annamaria, Procaccini, Enrica, Neri, Giandomenico, Vetere, Maurizio, Tomai, Fabrizio, Gaspardone, Achille, Biondi-Zoccai, Giuseppe, Schillaci, Orazio
• Format: Article
• Language: English
• Subjects: Aged; Cardiology; Female; Humans; Imaging; Male; maximal ischemia score; Medicine; Medicine & Public Health; Middle Aged; Myocardial ischemia; Myocardial Ischemia - diagnostic imaging; Myocardial Ischemia - mortality; myocardial perfusion imaging; Myocardial Perfusion Imaging - methods; myocardial perfusion scintigraphy; Nuclear Medicine; Original Article; Prognosis; Radiology; Retrospective Studies; Risk; segmentation
• ISSN: 1071-3581; 1532-6551
• DOI: 10.1007/s12350-014-9877-5

Myocardial perfusion scintigraphy (MPS) represents a key prognostic tool, but its predictive yield is far from perfect. We developed a novel clinically relevant segmentation method and a corresponding maximal ischemia score (MIS) in order to risk-stratify patients undergoing MPS. Patients referred for MPS were identified, excluding those with evidence of myocardial necrosis or prior revascularization. A seven-region segmentation approach was adopted for left ventricular myocardium, with a corresponding MIS distinguishing five groups (no, minimal, mild, moderate, or severe ischemia). The association between MIS and clinical events was assessed at 1 year and at long-term follow-up. A total of 8,714 patients were included, with a clinical follow-up of 31 ± 20 months. Unadjusted analyses showed that subjects with a higher MIS were significantly different for several baseline and test data, being older, having lower ejection fraction, and achieving lower workloads (P < .05 for all). Adverse outcomes were also more frequent in patients with higher levels of ischemia, including cardiac death, myocardial infarction (MI), and their composites (P < .05 for all). Differences in adverse events remained significant even after extensive multivariable adjustment (hazard ratio for each MIS increment = 1.57 [1.29-1.90], P < .001 for cardiac death; 1.19 [1.04-1.36], P = .013 for MI; 1.23 [1.09-1.39], P = .001 for cardiac death/MI). Our novel segmentation method and corresponding MIS efficiently yield satisfactory prognostic information.