Cardiac involvement in COVID-19 patients: mid-term follow up by cardiovascular magnetic resonance

Background Coronavirus disease 2019 (COVID-19) induces myocardial injury, either direct myocarditis or indirect injury due to systemic inflammatory response. Myocardial involvement has been proved to be one of the primary manifestations of COVID-19 infection, according to laboratory test, autopsy, and cardiovascular magnetic resonance (CMR). However, the middle-term outcome of cardiac involvement after the patients were discharged from the hospital is yet unknown. The present study aimed to evaluate mid-term cardiac sequelae in recovered COVID-19 patients by CMR Methods A total of 47 recovered COVID-19 patients were prospectively recruited and underwent CMR examination. The CMR protocol consisted of black blood fat-suppressed T2 weighted imaging, T2 star mapping, left ventricle (LV) cine imaging, pre- and post-contrast T1 mapping, and late gadolinium enhancement (LGE). LGE were assessed in mixed both recovered COVID-19 patients and healthy controls. The LV and right ventricle (RV) function and LV mass were assessed and compared with healthy controls. Results A total of 44 recovered COVID-19 patients and 31 healthy controls were studied. LGE was found in 13 (30%) of COVID-19 patients. All LGE lesions were located in the mid myocardium and/or sub-epicardium with a scattered distribution. Further analysis showed that LGE-positive patients had significantly decreased LV peak global circumferential strain (GCS), RV peak GCS, RV peak global longitudinal strain (GLS) as compared to non-LGE patients (p < 0.05), while no difference was found between the non-LGE patients and healthy controls. Conclusion Myocardium injury existed in 30% of COVID-19 patients. These patients have depressed LV GCS and peak RV strains at the 3-month follow-up. CMR can monitor the COVID-19-induced myocarditis progression, and CMR strain analysis is a sensitive tool to evaluate the recovery of LV and RV dysfunction.

such as cardiac stress due to respiratory failure, indirect injury from systemic inflammatory response-cytokine release syndrome, stress cardiomyopathy, or a combination of all these factors [6][7][8][9].
Cardiovascular magnetic resonance (CMR) can visualize and quantify heart volume and function and characterize the myocardial tissue; thus, it has been used as a gold standard non-invasive imaging tool in cardiovascular medicine [10]. A recent single-center study from Wuhan demonstrated that more than half of the recovered COVID-19 patients sustain cardiac edema, fibrosis, and impaired right ventricle (RV) contractile function [11]. However, in this small-sample retrospective study, only patients with reported cardiac symptoms were included. The middle-term outcome of cardiac involvement in COVID- 19 patients is yet unknown. Thus, the present study aimed to evaluate mid-term cardiac sequelae in recovered COVID-19 patients by CMR.
Age-and sex-matched healthy controls, who previously underwent CMR in our hospital were selected from a health screening database. All the controls had a normal electrocardiogram (ECG), echocardiography, and CMR and did not present any cardiovascular disease or systemic inflammation. The present study was approved by the local institutional review board (KS2020001), and informed consent was obtained from all patients.

CMR scanning protocol
All patients underwent CMR on a 3 T CMR scanner (Ingenia CX, Philips Healthcare, Best, The Netherlands). The CMR protocol consisted of black blood fat-suppressed T2 weighted imaging (T2w), T2 star mapping, left ventricle (LV) cine imaging including four chambers, two-chamber, short axis, pre-and post-contrast T1 mapping, and late gadolinium enhancement (LGE).

CMR images analysis
Both COVID-19 and healthy subject data were intermingled. And anonymized images were evaluated by 3 radiologists (ZZ, HW and LX with 5, 8 and 12 years of CMR diagnosis experience, respectively). The LGE lesion was quantified using full width at half-maximum method [12]. The visual presence and different patterns (epicardial, mid-wall, or transmural) on the LGE images were assessed by three radiologists independently. Any discrepancies were resolved by consensus. The ratio between the LGE volume and the total LV myocardium volume (LGE/myocardium) in the LGE-positive patients was calculated.
Whole LV myocardium region (including regions of LGE lesion) were delineated on the native T1 mapping and gobal T1 values were computed using cvi42 software (Circle Cardiovascular Imaging Inc., Calgary, Alberta, Canada).
The LV and RV function and LV mass were assessed based on the short-axis cine images using cvi42 software (Circle Cardiovascular Imaging). Endocardial and epicardial borders, with papillary muscles excluded from volumes, were identified automatically by the software and amended by a radiologist (HW). 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 (Circle Cardiovascular Imaging). 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 two radiologists (HW and ZZ respectively). Patients were further divided into two subgroups based on the presence of LGE.

Statistical analysis
All data were analyzed using SPSS (version 25.0, Statistical Package for the Social Sciences, International Business Machines, Inc., Armonk, New York, 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. χ 2 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.
Inter-and intra-observer intraclass coefficients (ICC) was used to assess reproducibility of strain measurements. The same investigator (HW) measured the 3D global strain of LV and RV in the all 75 studies on two occasions, separated by an interval of 2 months to assessment of intra-observer reproducibility. And a second investigator (ZZ) also evaluated all 75 subjects to assess inter-observer reproducibility.

Population characteristics
From May 8 to July 20, 2020, 47 patients who recovered from COVID-19 were recruited and underwent CMR. One patient was excluded because of moderate tricuspid regurgitation, one was excluded because of hypertrophic cardiomyopathy, and one was excluded because of hypertensive cardiomyopathy. The resulting 44 recovered patients and 31 healthy controls were included in this study. The clinical characteristics of COVID-19 patients are reported in Table 1. The average age of COVID-19 patients was 47.6 ± 13.3 years, and the cohort comprised of 19 (43.2%) men. Among the 31 normal controls, 19 were male, which matched with the COVID-19 group. Althoug only 12 female healthy controls were included, there was no significant difference in sex distribution among patients with LGE-positive, patients with LGEnegative and normal controls (P > 0.05)( Table 2).
Among all 44 recovered patients, normal ECGs were revealed in 43 patients. ST segment elevation in leads II, III, AVF and prolonged PR interval during initial diagnosis was demonstrated in one patient. Echocardiography was performed in 5 patients, all showed a small amount of mitral or/and tricuspid regurgitation.

LGE and T1 mapping
No myocardium hemorrhage was observed in any patient; however, LGE was detected in 13 patients (29.5%). All of the LGE-positive patients were from the COVID-19 group, no LGE was observed any of the healthy controls. All LGE lesions were located in the middle myocardium and/or sub-epicardium with a scattered distribution (Fig. 1). Case 1 was the only patient with abnormal ECG findings. Among a total of 208 myocardial segments in 13 patients, most LGE lesions were located at the inferior and inferior-lateral segments at the base and mid-chamber. The Bull's eye illustration (Fig. 2) shows us the number of myocardial LGE distributed in the American Heart Association 16 segments' model for all 13 patients. The inferior wall and inferior-lateral wall of the basal segment was the most frequently involved segment (10/12 patients). The median of LGE/myocardium ratio was 1.7% (1.1-3.0%).
Global native T1 showed no significant difference between COVID-19 patients with and without LGE (1286 ms ± 60 ms vs 1253 ms ± 55 ms, P > 0.05). Table 2 shows the values for global LV and RV morphological and functional parameters and the measurement of the 3D global CMR feature-tracking deformation parameters. No significant difference was detected in the LV and RV morphological parameters (EDV, ESV, and mass) among COVID-19 patients with and without LGE and normal controls. Although no significant difference was detected in the LV and RV traditional function parameters-EF, CO, cardiac index (CI), SV; LV peak GCS was decreased in COVID-19 patients with visual LGE (− 15.1 ± 10.3) on CMR images as compared to healthy controls (− 19.4 ± 3.0) (p < 0.05). Both peak 3D GCS and GLS of RV in COVID-19 patients with LGE (− 9.4 ± 3.4 and − 7.8 ± 4.0, respectively) were significantly decreased as compared to COVID-19 patients without LGE (− 12.1 ± 4.0 and − 12.9 ± 3.0, respectively) and normal control (− 12.9 ± 4.3 and − 11.3 ± 3.9, respectively) (both p < 0.05). However, no difference (peak 3D GCS and GLS of RV) was detected between COVID-19 patients without LGE and the normal controls (p > 0.05).

Discussion
The present prospective study among recovered COVID-19 patients found that ~ 30% had CMR evidence for myocardial injury (manifested as LGE) at 3-month follow up.
LGE-positive patients had a decrease in the LV GCS compared to healthy controls and a significant decrease in the RV GCS and GLS as compared to non-LGE COVID-19 patients.
Myocardial involvement has been proved to be one of the primary manifestations of COVID-19 infection by laboratory, autopsy, and CMR [14][15][16] as compared to other members of the coronavirus family [17]. A recent study demonstrated that in the recovered COVID-19 patients with cardiac symptoms, 54% had myocardium edema, and 31% had LGE [11]. The persistence of visual LGE at 3-month follow-up reflecting necrosis or scar (fibrosis) might be caused by COVID-19 and needs to be elucidated further [18]. Yet, visual LGE indicated that up to 30% COVID-19 patients have irreversible myocardial injury [19,20], which is consistent with the findings of the previous study and can confirm one of the main mechanisms of COVID-19 induced direct myocardial injury-myocarditis [21][22][23]. In addition, the presence of LGE has been proven to be an independent predictor of all-cause mortality and cardiac mortality in myocarditis [24,25]. Since the LGE/myocardium ratio was small, the cardiac status of COVID-19 patients with LGE needs to be closely monitored.
We found all of the LGE lesions to be located in the middle myocardium and/or sub-epicardium, wherein the fibers are oriented transversely, and the torque enhances shortening in the circumferential direction [26,27]. This may be the reason for LV 3D GCS to be oriented along the perimeter in short axis view [26] and decreased in the COVID-19 patients with LGE. This finding prompts us to focus on the LV circumferential contraction dysfunction in these patients in addition to impaired RV function.
Impaired RV function in COVID-19 survivors has been demonstrated by echocardiography and CMR [11,28,29]. RV strain has been recommended to assess the RV function in clinical scenarios with suspected RV dysfunction [30]. Although our results showed that the RV traditional morphological and function parameters are in a normal range, the RV strain in COVID-19 patients with LGE was significantly decreased as compared to those without LGE and healthy controls. This phenomenon indicated that COVID-19 patients with LGE still had RV dysfunction, which could be detected by CMR strain analysis. COVID-19 results in acute respiratory distress syndrome (ARDS) [31] and is frequently associated with RV dysfunction, increased pulmonary resistance [32], increased values of systolic pulmonary arterial pressure, increased RV afterload [33,34], severe hypoxia, oxidative stress, and increased myocardial oxygen demand induced by ARDS [35].
According to a recent systematic echocardiographic study, the most frequent abnormality induced by COVID-19 was RV dilation with or without dysfunction [32]. However, our results only showed RV dysfunction in patients with LGE. No RV morphological (EDV and ESV) difference was detected at the 3-month follow up, which might be attributed to improved pneumonia; subsequently, the RV rapidly returns to normal size while the RV strain is still decreased. Global native T1 value in both LGE-positive and LGE-negative COVID-19 patients were elevated compared with [1122 (1100, 1143) ms and 1122 ± 57 ms] healthy subjects, performed at Philips 3 T CMR with MOLLI sequence, previously reported by Gottbrecht et al. [36] and Clotilde Roy et al [37].

Limitations
The current study has limitations. First, the lack of baseline pre-COVID-19 CMR examination limits the evaluation of the progress of heart involvement. Second, the sample size was relatively small. Third, the majority of the recruited patients had moderate and severe COVID-19, and hence, our study does not reflect the full spectrum of critical COVID-19 patients. Fourth, we had data of only a 3-month CMR examination, and thus, a longtime follow-up is essential to determine the progression or regression of cardiac involvement. Fifth, we did not have native T1 of the healthy control group, which would be useful information to compare with the COVID-19 patients. However, we provide reference normal T1 value base on the previous studies. Sixth, we did not have information regarding hemodynamic instability or evidence of cardiac involvement during initial diagnosis or hospitalization on all patients, which would be helpful to correlate those kinds of patients with the CMR findings.

Conclusions
Myocardium injury is present in nearly a third of COVID-19 patients at 3 months with evidence of impaired LV GCS and RV GCS and GLS. CMR may be useful to monitor the COVID-19-induced myocarditis progress and remains to be examined.