Coronary artery disease modifies the association of electrocardiography parameters with myocardial structure and function

Background: Although there are many publications separately describing ECG abnormalities in cardiac disease, published data on electromechanical associations with subclinical coronary artery disease (CAD) are limited. We sought to investigate the relationship of ECG variables and CMR defined tissue structure in adults with and without subclinical CAD.

Methods: We prospectively included 173 adults without heart failure who had history of SARS-CoV-2 infection (46% men, mean age 55 ± 13 years). Participants underwent electrocardiography, ultra-high-resolution coronary CT (UHR-CT), and 3T contrast-enhanced CMR. CAD+ was defined for presence of atherosclerotic plaque in any coronary arteries on UHR-CT. Left and right ventricular (LV and RV) and left atrial (LA) function and structure were measured from CINE images. Myocardia T1 and T2 were defined by T1 and T2 mapping respectively. Multivariable analysis stratified by the presence of CAD adjusted for demographics was used to examine associations between ECG and CMR variables (P < 0.05 for statistical significance).

Results: 72 participants (CAD+, 42%) were found to have CAD (Table 1). CAD+ participants were older (P < 0.001) and more commonly men (P = 0.001) with 3% higher native myocardial T1 (P = 0.004). Longer QRS duration was associated with greater ventricular volumes, stoke volume, and LV mass (all P < 0.05) in both groups with higher strength of association in the CAD+. However, QRS duration was also associated with LA volume, strain, and myocardial T1 and T2 in CAD‒ (all P < 0.05). PR and QT intervals were associated with LV and RV volumes, stroke volumes, and LV mass in CAD+. Additionally, PR interval was associated with LV strain and QT interval was associated with LV end-systolic volume and LV mass in CAD‒ (all P < 0.05). Figure 2 displays the correlation between ECG and CMR variables dichotomously and example scatter plots.

Conclusion: In this community-based cohort with history of COVID-19 infection, we characterized the associations between electrocardiographic values and CMR defined myocardial characteristics as measured by T1 and T2 mapping. Principle findings include 1) In adults with CAD, PR and QT intervals were associated with both LV and RV structure and function. PR interval was also associated with LA volume. 2) In adults without CAD, PR and QT intervals were only associated with LV strain and myocardial structure, respectively. QRS duration was associated with LA volume, strain, and myocardial T1 and T2. 3) QRS duration was associated with both LV and RV structure and function regardless of CAD status with greater strength of association in adults with CAD. These findings demonstrate that summation electrograms as in the routinely used 12 lead electrocardiogram reflect cardiac morphology and function, as well as myocardial T1 and T2 parameters. Moreover, such associations are modified by the presence or absence of coronary atherosclerosis defined in detail by UHR-CT angiography.

Table 1. Coronary artery disease (CAD) status-stratified demographics, ECG, and CMR data of the study cohort. Values presented as mean ± SD or as N (%). P values compare between groups with and without CAD. Bold values indicate P < 0.05.
Figure 1. Representative images from a study participant. CMR measured ejection fraction was 62%. The ECG (left) shows sinus bradycardia (heart rate 47 bpm), and right bundle branch block (QRS duration 136 ms) with 4.6% LGE (middle) and extensive atherosclerotic disease (right).
Figure 2. Summary of relationship between ECG and CMR variables. ×: no correlation, ˅: with correlation in the multivariable analysis (P < 0.05). Regression plots of QRS vs LV EDV (end-diastolic volume) and LV mass according to the coronary artery disease (CAD) status. QRS was positively correlated with LV EDV and mass, which indicated prolonged QRS duration with greater LV EDV and mass. The slopes are higher in the group with CAD. R and P values were from the univariable analysis.