Prevalence and prognosis of unrecognized myocardial infarction determined by cardiac magnetic resonance in older adults

Abstract

Context: Unrecognized myocardial infarction (MI) is prognostically important. Electrocardiography (ECG) has limited sensitivity for detecting unrecognized MI (UMI).

Objective: Determine prevalence and mortality risk for UMI detected by cardiac magnetic resonance (CMR) imaging or ECG among older individuals.

Design, setting, and participants: ICELAND MI is a cohort substudy of the Age, Gene/Environment Susceptibility-Reykjavik Study (enrollment January 2004-January 2007) using ECG or CMR to detect UMI. From a community-dwelling cohort of older individuals in Iceland, data for 936 participants aged 67 to 93 years were analyzed, including 670 who were randomly selected and 266 with diabetes.

Main outcome measures: Prevalence and mortality of MI through September 1, 2011. Results reported with 95% confidence limits and net reclassification improvement (NRI).

Results: Of 936 participants, 91 had recognized MI (RMI) (9.7%; 95% CI, 8% to 12%), and 157 had UMI detected by CMR (17%; 95% CI, 14% to 19%), which was more prevalent than the 46 UMI detected by ECG (5%; 95% CI, 4% to 6%; P < .001). Participants with diabetes (n = 337) had more UMI detected by CMR than by ECG (n = 72; 21%; 95% CI, 17% to 26%, vs n = 15; 4%; 95% CI, 2% to 7%; P < .001). Unrecognized MI by CMR was associated with atherosclerosis risk factors, coronary calcium, coronary revascularization, and peripheral vascular disease. Over a median of 6.4 years, 30 of 91 participants (33%; 95% CI, 23% to 43%) with RMI died, and 44 of 157 participants (28%; 95% CI, 21% to 35%) with UMI died, both higher rates than the 119 of 688 participants (17%; 95% CI, 15% to 20%) with no MI who died. Unrecognized MI by CMR improved risk stratification for mortality over RMI (NRI, 0.34; 95% CI, 0.16 to 0.53). Adjusting for age, sex, diabetes, and RMI, UMI by CMR remained associated with mortality (hazard ratio [HR], 1.45; 95% CI, 1.02 to 2.06, absolute risk increase [ARI], 8%) and significantly improved risk stratification for mortality (NRI, 0.16; 95% CI, 0.01 to 0.31), but UMI by ECG did not (HR, 0.88; 95% CI, 0.45 to 1.73; ARI, -2%; NRI, -0.05; 95% CI, -0.17 to 0.05). Compared with those with RMI, participants with UMI by CMR used cardiac medications such as statins less often (36%; 95% CI, 28% to 43%, or 56/157, vs 73%; 95% CI, 63% to 82%, or 66/91; P < .001).

Conclusions: In a community-based cohort of older individuals, the prevalence of UMI by CMR was higher than the prevalence of RMI and was associated with increased mortality risk. In contrast, UMI by ECG prevalence was lower than that of RMI and was not associated with increased mortality risk.

Trial registration: clinicaltrials.gov Identifier: NCT01322568.

Figure 1. CMR examples of recognized MI (A), no MI (B), and unrecognized MI (C and D)

Panel A demonstrates a recognized myocardial infarction involving the anterior and anteroseptal segments in the typical left anterior descending artery distribution as seen on late gadolinium enhancement imaging (arrow) with the corresponding diastolic cine frame on the right. Panel B demonstrates a participant with no evidence of myocardial infarction on late gadolinium enhancement imaging. The myocardium is uniformly dark (“nulled”) on the late gadolinium enhancement image (left). Panel C demonstrates an unrecognized myocardial infarction in the basal inferolateral wall on late gadolinium enhancement imaging (arrow) with the corresponding end-diastolic cine frame on the right. Panel D demonstrates two unrecognized myocardial infarctions in different coronary territories in the same participant. There is a small myocardial infarction in the inferolateral wall (arrow) corresponding to the left circumflex artery territory and a larger myocardial infarction involving the anterior and anteroseptal segments (arrow) corresponding to left anterior descending artery territory. The corresponding end-diastolic cine frame is shown on the right.

Figure 2. Mortality curves according to myocardial infarction status

The mortality was similar (p=0.399) between recognized and unrecognized MI, and the mortality was significantly worse (p<0.001) for those with unrecognized MI versus those without MI based on the log rank test.