Background: older patients with ST-segment elevation myocardial infarction (STEMI) represent a very high-risk population. Data on the prognostic value of cardiac magnetic resonance (CMR) in this scenario are scarce.
Methods: the registry comprised 247 STEMI patients over 70 years of age treated with percutaneous intervention and included in a multicenter registry. Baseline characteristics, echocardiographic parameters and CMR-derived left ventricular ejection fraction (LVEF, %), infarct size (% of left ventricular mass) and microvascular obstruction (MVO, number of segments) were prospectively collected. The additional prognostic power of CMR was assessed using adjusted C-statistic, net reclassification index (NRI) and integrated discrimination improvement index (IDI).
Results: during a 4.8-year mean follow-up, the number of first major adverse cardiac events (MACE) was 66 (26.7%): 27 all-cause deaths and 39 re-admissions for acute heart failure. Predictors of MACE were GRACE score (HR 1.03 [1.02-1.04], P < 0.001), CMR-LVEF (HR 0.97 [0.95-0.99] per percent increase, P = 0.006) and MVO (HR 1.24 [1.09-1.4] per segment, P = 0.001). Adding CMR data significantly improved MACE prediction compared to the model with baseline and echocardiographic characteristics (C-statistic 0.759 [0.694-0.824] vs. 0.685 [0.613-0.756], NRI = 0.6, IDI = 0.08, P < 0.001). The best cut-offs for independent variables were GRACE score > 155, LVEF < 40% and MVO ≥ 2 segments. A simple score (0, 1, 2, 3) based on the number of altered factors accurately predicted the MACE per 100 person-years: 0.78, 5.53, 11.51 and 78.79, respectively (P < 0.001).
Conclusions: CMR data contribute valuable prognostic information in older patients submitted to undergo CMR soon after STEMI. The Older-STEMI-CMR score should be externally validated.
Keywords: cardiac magnetic resonance; myocardial infarction; older patients; older people; prognosis; risk.
© The Author(s) 2022. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For permissions, please email: journals.permissions@oup.com.