Global longitudinal strain to identify low-risk patients with suspected ACS

Abstract

Abstract Funding Acknowledgements Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): Albert Einstein Society Background Determining which chest pain presentations should be treated and considered an acute coronary syndrome/myocardial infarction (ACS/MI) versus those with non-cardiac chest pain (NCCP) can be difficult. Initial evaluation of suspected ACS requires assessment of presenting symptoms, risk factors, ECG, and cardiac biomarkers. Bedside echocardiography can assist in rapid assessment of suspected ACS through measurement of echocardiographic wall motion score index and left ventricular ejection fraction, though the sensitivity of these measurements has been called into question. Global longitudinal strain (GLS) has been associated with significant CAD and has been found to be more reproducible than LVEF. However, its utility in rapid ED evaluation of chest pain remains under-explored. Purpose Assess the utility of speckle-tracking strain in addition to clinical and demographic factors in identification of low-risk patients among those presenting to the ED with suspected ACS. Methods This was a retrospective single center study of 434 hospitalized patients aged 18 years or older in whom ACS (excluding STEMI) was suspected by ED assessment, from 9/1/2015 – 12/31/2019. Echocardiography within 24 hours of admission was analyzed, with left ventricular global longitudinal strain (LVGLS) obtained via AutoSTRAIN software (TOMTEC Imaging Systems GmbH). Patients were identified as having NCCP (n = 158, 36%), myocardial injury (n = 110, 25%), or MI (n = 166, 38%; subdivided into NSTEMI [n = 74, 44.6%] and type II MI [n = 92, 55.4%]) according to the 4th universal definition of MI. Mean strain values were compared between study groups using Independent T tests. Logistic regression and ROC analysis was done to determine the value of LVGLS in the prediction of ACS. Results Non-white subjects were over-represented in the NCCP group (92% vs 8%), versus the myocardial injury and MI groups (65% vs 35%, p < 0.001), and on average the NCCP group was younger (56.5 ± 14.5 vs 64.8 ± 15, p < 0.001). LVGLS was significantly higher for NCCP versus the MI group (17.7 ± 2.8 vs 14.9 ± 3.9, p < 0.001). ROC analysis (c-statistic = 0.72) identified an optimal cutoff at ≤15.6, with sensitivity of 56% and specificity of 82%. Logistic regression analysis, including demographic and clinical variables, identified age, LVGLS, LV end-diastolic volume and serum creatinine as significant independent predictors for NCCP vs ACS. The addition of these factors in the predictive analysis resulted in slightly improved model performance (c-statistic = 0.78). Conclusions LVGLS among patients with suspected ACS is significantly different between NCCP and MI; however, low sensitivity for MI makes it inadequate as a single test to discriminate between the two. Combining LVGLS with other clinical/laboratory factors may have potential utility and will be explored in future work. Abstract Figure. Distribution Plot for LVGLS Abstract Figure. ROC Curve for LVGLS