Prediction of Coronary Artery Spasm in Patients Without Obstructive Coronary Artery Disease Using a Comprehensive...

Background: The lack of an accurate coronary artery spasm (CAS) risk prediction model highlights the failure to consider dynamic coronary health and reveals a gap in understanding CAS. Methods: A total of 913 Taiwanese patients (460 women and 453 men) with suspected ischemic heart disease but without angiographic obstructive coronary artery disease were subjected to intracoronary methylergonovine testing during the period 2008-2025. Results: The study included 645 CAS cases (70.6%) and 268 non-CAS controls (29.4%). The multivariable logistic regression model identified 10 variables significantly associated with CAS (p < 0.05): male sex, smoking, low systolic and diastolic blood pressure, reduced B-type natriuretic peptide levels, elevated low-density lipoprotein levels, increased relative wall thickness at end-systole, high left ventricular mass index, low e'(l) values, and high Tei index. Discrimination performance was moderate, with an AUC value of 73.8% that dropped to 72.4% after bootstrapped internal validation, suggesting the potential generalizability of the derived model. The total score ranged from 36 to 98, representing a predicted probability between 12% and 98%, respectively. Conclusions: While a total score of ≥58 with the probability of CAS exceeding 50% indicates a significant chance of undiagnosed CAS, for patients with a total score ≥ 69 and a high probability of CAS ≥ 75%, coronary catheterization with CAS provocation testing is strongly recommended for a definite diagnosis. The simple 10-variable scoring model allows ranking of at-risk populations and is designed to be used as a screening tool rather than a diagnostic adjunct, enabling more efficient diagnostic resource allocation.