Study design and participants
For this prospective, single-center study, we recruited consecutive COVID-19 patients from May 8 to July 20, 2020. The inclusion criteria were as follows: (1) Patients were previously confirmed to have SARS-CoV-2 infection, (2) patients were recovered from COVID-19 and discharged from the hospital for 12 weeks, (3) patients agreed to participate in the study and signed informed consent. The exclusion criteria were as follows: (1) Patients has undergone pacemaker surgery, (2) patients with uncontrolled high blood pressure, (3) patients with coronary heart disease (evidence of coronary artery stenosis > 50%) or previous myocardial infarction, (4) patients with moderate to severe valvular dysfunction, (5) patients with previous atrial fibrillation, (6) previous heart failure, (7) previous myocarditis, (8) patients with known cardiomyopathy, (9) patients with severe renal insufficiency (creatinine clearance rate < 30 mL/min/1.73 m2, (9) patients who cannot cooperate with breath-holding and cannot undergo CMR examination, (10) pregnant women, (11) patients are not suitable as clinical subjects due to other factors.
Age- and sex-matched healthy controls, who underwent the cardiac MRI exams in our hospital previously, were selected from a health screening database. All the controls showed normal ECG, echocardiography, and cardiac MRI and did not present any cardiovascular disease or systemic inflammation. The present study was approved by the local institutional review board hospital (KS2020001), and informed consent was obtained from all patients.
MRI scanning protocol
All patients underwent MRI examinations on a 3T MR scanner (Ingenia CX, Philips Healthcare, Best, The Netherlands). The cardiac MRI protocol consisted of black blood fat-suppressed T2 weighted imaging (BB-T2WI), T2 star mapping, left ventricle cine imaging including four chambers, two-chamber, short axis, pre- and post-contrast T1 mapping, and late gadolinium enhancement (LGE).
BB-T2WI was performed in short axis using multishot turbo spin echo (TSE) sequence with time of repetition (TR) = 2 heart beat periods, time of echo (TE) = 75 ms, voxel size = 1.7 × 1.7 × 8 mm3, field of view (FOV) = 380 × 380 mm2, flip angle (FA) = 90°, and acceleration factor = 2. T2 star mapping was carried out by a turbo field echo (TFE) sequence with 15 TEs from 1.15 ms to 16 ms, TR = 29 ms, voxel size = 1.8 × 1.8 × 15 mm3, FOV = 300 × 300 mm2, FA = 25°, and acceleration factor = 3. The cine scanning was conducted by a balanced steady state-free precession (bSSFP) sequence, with TR/TE = 3.0/1.52 ms, FA = 45°, voxel size = 1.8 × 1.8 × 8 mm3, FOV = 270 × 270 mm2, and acceleration factor = 1.5 or 3.6 depending on the number of slices acquired per breath hold. For short axis cine imaging, 9 slices were acquired to cover the entire left ventricle. Modified Look-locker inversion recovery (MOLLI) acquisition scheme was applied for both pre- and post-contrast T1 mapping using single-shot bSSFP sequence with TR/TE = 3.3/1.5 ms, voxel size = 2 × 2 × 15 mm3, FOV = 300 × 300 mm2, FA = 20°, and acceleration factor = 2. LGE imaging was implemented 10 min after intravenous administration of contrast medium (0.2 mmol/kg; Magnevist gadopentetate dimeglumine, Bayer Healthcare, Bayer, Bergkamen, Germany) using a phase-sensitive inversion-recovery (PSIR) TFE sequence with TR/TE = 6.1/3.0, voxel size = 1.6 × 1.9 × 8 mm3, FOV = 350 × 350 mm2, FA = 25°/5°, and acceleration factor = 2. All the protocols were executed using ECG trigger and breath holding.
MRI images analysis
Anonymized images were evaluated by two radiologists (HW and LX with 8 and 12 years of cardiac MRI diagnosis experience, respectively). Myocardium edema was defined as the regional or global signal hyperintensity on T2WI13. Myocardial edema ratio (ER) was defined as the ratio > 1.9 between myocardial signal intensity (SI) and skeletal muscle SI. High signal area of inadequately suppressed slow-flowing cavitary blood was excluded carefully14. The LGE lesion was quantified using full width at half-maximum method15. The visual presence and different patterns (epicardial, mid-wall, or transmural) on the LGE images were assessed by two radiologists independently. Any discrepancies were resolved by reaching consensus through consultation. The ratio between the LGE volume and the total LV myocardium volume (LGE/myocardium) in the LGE-positive patients was calculated.
The left ventricle (LV) and RV function and LV mass were assessed based on the short-axis cine images using cvi42 software (Circle Cardiovascular Imaging Inc., Calgary, Canada). Endocardial and epicardial borders, with papillary muscles excluded from volumes, were identified automatically by the software and amended by a radiologist (WH). LV and RV range were defined from the planes of the mitral valve and tricuspid valve to the apex, respectively, on 4-chamber cine images in both diastolic and systolic phases. Short-axis images were divided into size-based equiangular segments with RV-LV junction as the reference point. LV and RV function parameters, end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV), cardiac output (CO), ejection fraction (EF), and LV mass were calculated automatically. All volumes and masses were normalized to the body surface area (BSA).
Three-dimensional (3D) global radial strain (GRS), global circumferential strain (GCS), and global longitudinal strain (GLS) of LV and RV were obtained using cvi42. The end-diastolic phase served as the reference. Contours of endo- and epicardial myocardium of short-axis, as well as the 2-, 3-, 4-chamber long-axis cine images, were drawn by a radiologist (WH). Patients were further divided into two subgroups based on visual LGE.
Statistical analysis
All data were analyzed using SPSS software version 25.0 (SPSS Inc., Chicago, IL, USA). Normally distributed continuous variables were expressed as mean ± standard deviation. Two-tailed one-way ANOVA was used to analyze the differences between LGE, non-LGE, and normal control groups. Categorical variables were expressed as counts and percentages. T-test was used to compare the means with normal distribution, and the Mann–Whitney U test was used to compare the variables with non-normal distribution between LGE and non-LGE groups. χ2 test was used to explore the statistical significance of CMR parameters among LGE, non-LGE, and normal control groups. A two-sided p < 0.05 was considered as statistically significant.