CMR Findings in COVID-19 Recovered Patients: A Review on Parametric Mapping, Feature-Tracking, and LGE

On March 11, 2020, the World Health Organization raised the coronavirus disease 2019 (COVID-19) status to a pandemic level. The disease caused a global outbreak with devastating consequences, and a fair percentage of patients who have recovered from it continue experiencing persistent sequelae. Hence, identifying the medium and long-term effects of the COVID-19 disease is crucial for its future management. In particular, cardiac complications, from affected function to myocardial injuries, have been reported in these patients. Considering that cardiovascular magnetic resonance (CMR) imaging is the gold standard in diagnosing myocardial involvement and has more advantages than other medical imaging modalities, assessing the outcomes of patients who recovered from COVID-19 with CMR could prove beneficial. This review compiles common findings in CMR in patients from the general population who recovered from COVID-19. The CMR-based techniques comprised parametric mapping for analyzing myocardial composition, feature tracking for studying regional heart deformation, and late gadolinium enhancement for detecting compromised areas in the cardiac muscle. A total of 19 studies were included. The evidence suggests that it is more likely to find signs of myocardial injury in patients who recovered from COVID-19 than in healthy controls, including changes in T1 and T2 mapping relaxation times, affected strain, or the presence of late gadolinium enhancement (LGE) lesions. However, more than two years after the outbreak, there is still a lack of consensus about how these parameters may indicate cardiac involvement in patients who recovered from the disease, as limited and contradictory data is available.


Introduction
Coronavirus disease 2019 (COVID-19) is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that reached pandemic levels in March 2020.Up to July 22, 2022, the number of total cases compiled by the Center for Systems Science and Engineering at John Hopkins University reported 567,951,340 million cases and 6,380,835 deaths worldwide [1].COVID-19 manifests mainly through respiratory symptoms.However, most organs are affected by the disease, including the cardiovascular system [2][3][4].People infected with COVID-19 have a greater risk of experiencing cardiovascular disease, regardless of the disease severity and vaccination status [5].They also are likely to suffer from myocarditis and myocardial injury [6][7][8][9][10].
Imaging modalities are fundamental in diagnosis, especially since COVID-19 requires a prompt response.For likely cardiovascular involvement, bedside echocardiography should be used as the first step for diagnosis and further referring [11].Computed tomography might help identify pneumonia and rule out suspected causes of cardiac dam-age [11].Nuclear medicine imaging could help diagnose pulmonary embolus, but its more comprehensive benefit is limited [11,12].On the other hand, cardiovascular magnetic resonance (CMR) has several advantages in assessing myocardial tissue and is the gold standard in diagnosing myocardial involvement [11,13].
Besides being non-invasive, CMR encompasses advanced techniques that provide qualitative and quantitative information about cardiac function.Three of them are parametric mapping, feature-tracking (CMR-FT), and late gadolinium enhancement (LGE).CMR parametric mapping allows a quantitative analysis of regional myocardial composition based on changes in the relaxation times of water protons in the tissue (T1, T2, and T2*) and the extracellular volume (ECV) [14], aiding the quantification of myocardial disease processes.On the other hand, CMR-FT allows a quantitative analysis of regional heart deformation by myocardial strain assessment [15,16].Finally, LGE detects compromised areas in the cardiac muscle depending on the distribution of a contrast medium in the extracellular or intravascular space [17].Two years after the COVID-19 pandemic began, the cardiovascular impact of this disease is better known.This work critically reviews the most relevant CMR imaging findings in parametric mapping, myocardial strain, and LGE in recovered COVID-19 patients from the general population.

Methods
For this literature review, we performed a comprehensive literature review in PubMed, Scopus, and Google Scholar, including the keywords "SARS-CoV-2" or "COVID-19" and "CMR" or "MRI" or "cardiac MR" and "MAPPING" or "LGE" or "T1" or "T2" or "STRAIN" or "FEATURE" or "TRACKING" or "CMR-FT".We considered documents available until May 24, 2022.The search was limited to publications from and including 2020.We excluded case reports, reviews, editorials, comments, preprints, and documents in a different language than English.We selected scientific articles focused on adults.The search in PubMed and Scopus was done in R [18] (version 4.1.2;RStudio 2021.09.1 build 372, PBC, Boston, USA) using the "easyPubMed" and "rscopus" packages.The Google Scholar search was done manually, and the records were exported with their built-in tool.All the steps in the screening stage were performed in R. Duplications were removed based on the record's digital object identifier (DOI).
The study eligibility was evaluated by verifying it contained reported CMR data on patients recovered from COVID-19, with a minimum number of subjects equal to 15 and at least 30 days from the diagnosis to the CMR study.Similarly, the type of study included retrospective, prospective, case-control, and research letters, provided they complied with the other criteria.We excluded reports considering specific populations, i.e., athletes.
The effect size pooling between controls and recovered patients was determined with a random-effects model using standardized mean differences.This analysis included healthy volunteers and excluded other comparison groups.The confidence interval around the pooled effect was determined with Knapp-Hartung adjustments [19].The heterogeneity was determined with Higgins & Thompson's I 2 statistic derived from Cochran's Q [20] and the heterogeneity variance τ 2 with the restricted maximum-likelihood estimator [21].Outliers were identified and removed, and influence diagnostics were also performed.The analyses were performed using the R packages meta, dmetar, and metafor.For the studies reporting median (interquartile range) values, those were transformed to mean (standard deviation) using the Wan et al. [22] method.
The total number of included subjects was 2007, from which 1217 were patients who recovered from COVID-19, 502 were controls or healthy volunteers, and 288 were from other comparison groups, i.e., patients with myocarditis or suspicion of myocardial injury.The mean age of participants was 45.9 ± 13.1 years (recovered patients 47.2 ± 13.5 years, controls 44.8 ± 12.9 years, other comparison groups 43.5 ± 14.0 years).Table 1 summarizes cohort details in each study.

CMR Parameters
COVID-19 has been linked to myocardial inflammation and myocardial injury [7,8,42] following the established CMR criteria for such a diagnosis.The updated Lake Louis criteria include parametric mapping for diagnosing myocardial inflammation: while native T1 mapping and ECV are linked to myocardial injury, T2 mapping is linked to myocardial edema [43].Extensive works about the connection between COVID-19 and myocarditis can be found elsewhere [6,7,9,42,44,45] and not be described here.
In other studies, patients who recovered from COVID-19 had lower LVEF and LVEDV than patients with non-COVID-19 myocarditis [26] and LVEF and myocardial mass than risk-factor matched patients; a higher LVEDVI was reported for this subgroup [35].Similarly, lower LVEF and LVSVI were found in patients who recovered from COVID-19 and had multisystem inflammatory syndrome (MIS)-myocarditis than those with non-MIS myocarditis.RVEF was also significantly lower in recovered subjects compared to historical control [28] and risk-factor matched groups [35].Finally, patients with acute non-COVID-19 related myocarditis were found to have a higher RVESVI than patients with persistent cardiac symptoms after a COVID-19 infection [36].
Studies including patients who recovered from COVID-19 reported increased [4,27,28,33,35,36], slightly increased [29,32,38,39,41] or similar [31,34] T1 native mapping values compared to controls.In particular, Kotecha et al. [28], Puntmann et al. [35], and Thornton et al. [37] found significantly higher T1 native mapping values in large groups of recovered subjects (148, 100 and 90, respectively) compared to different control groups, including patients without myocardial injury, healthy volunteers or controls, and risk factor-matched controls [26,28,35,37].In addition, patients who recovered from COVID-19 and had MIS-myocarditis had higher T1 native values than those with non-MIS myocarditis [30].Other studies assessing slightly higher or similar T1 native values did not reach statistical significance [29,31,32,34,38,39].The pooled effect size for T1 native values from studies reporting values measured at 1.5 T was 0.59 (95% CI 0.25 to 0.94) and was statistically significant (p = 0.0054), with moderate heterogeneity.On the other hand, the equivalent pooled effect size resulting from studies measuring at 3.0 T was 1.96 (95% CI 0.06 to 3.86) and significant (p = 0.0452), with high between-study heterogeneity.This result means a significant difference between the T1 native values of recovered patients and controls in all studies, favoring increased values for the patients.The corresponding forest plots are shown in Supplementary Fig. 1.
Generally, T2 values result higher with increased water content in the myocardium, a feature of both ischemic and non-ischemic cardiomyopathies commonly associated with acute myocardial inflammation [14,48].With image quality and reproducibility as the limiting factors of its wider clinical adoption, T2 mapping emerged with its quantitative nature and higher robustness [14].Though in the case of myocarditis, T1 mapping has better diagnostic accuracy and positive and negative predictive values [49,50], T2 mapping seems superior for assessing this disease activity in patients [51].Therefore, in patients with increased T2 values, an active inflammatory process is expected.In the case of lower T2 values [28], these have been reported in healthy males compared to females [52], with an unknown underlying reason for this phenomenon.

Cardiovascular Magnetic Resonance -Feature Tracking (CMR-FT)
CMR-FT is an emerging tool for quantitative analysis of regional heart deformation [15,16].It is based on optical flow, a technique used to track the movement of individual pixels in a series of images [16].Such a process allows contouring different heart regions to follow their movement and measure parameters related to their deformation, which can offer insight into underlying cardiac problems.One of those parameters is the strain, which, in the LV, describes regional changes related to the shortening, thickening, or lengthening of the myocardium.The assessment is usually done through the GCS, GRS, and GLS.Several studies have found that strain may serve as a more efficient marker of contractile dysfunction than other clinical ones [53][54][55].
CMR-FT assessment in patients who recovered from COVID-19 showed that their left ventricle global longitudinal strain (LVGLS) is lower than controls [31,32,36,41].One study reported similar values for patients who recovered from COVID-19 without LGE and controls and significantly lower values in the case of patients with LGE [39].Regarding the left ventricle global radial strain (LVGRS), no differences were found in any of the selected studies that reported this parameter [31,38,39,41].Left ventricle global circumferential strain (LVGCS), on the other hand, compared to healthy controls, was significantly lower in recovered patients with LGE [39], the delta variant [41], or acute non-COVID-19-related myocarditis [36].One study reported that the right ventricle global circumferential strain (RVGCS) and the right ventricle global longitudinal strain (RVGLS) were significantly lower in recovered patients with LGE than those without LGE and healthy controls, and the right ventricle global radial strain (RVGRS) had no changes [39].Finally, RVGLS was also lower in patients who recovered from COVID-19 compared to controls but higher than in patients with acute non-COVID-19-related myocarditis [36].The data summary of LV and RV strains is shown in Supplementary Table 1.

Late Gadolinium Enhancement (LGE)
LGE is the gold standard and most validated technique for assessing myocardial scar, inflammation, or necrosis [17,56].This technique allows differentiating normal and abnormal myocardium based on their T1 longitudinal relaxation times using an extracellular gadolinium-based contrast agent.In normal tissue, the cell membrane is compact, and the contrast agent is washed out quickly, resulting in a low concentration of gadolinium and a longer T1.In abnormal tissue, the contrast agent accumulates, shortening T1 [57,58].Therefore, affected areas appear hyperintense in an LGE-CMR image depending on the tissue's physiological properties.
LGE in the myocardium has been well documented as a negative predictive factor in many cardiac conditions, such as dilated and hypertrophic cardiomyopathy [63].It has also been linked to higher mortality in cardiac amyloidosis [64].Gutman et al. [65] showed that including LGE assessment by CMR can lead to a better selection of patients with an indication for implantable cardioverter-defibrillator implantation.
On the other hand, although the general prognosis of patients with myocarditis seems rather good [66], it tends to have a very variable course ranging from complete remission to severe complications [67].Some studies have also shown a worse prognosis in patients with positive LGE in myocarditis or myocardial inflammation [68].However, the data are limited, and persistent LGE was reported in over 50% of these patients in a one-year CMR follow-up [66].
While no long-term follow-up of COVID-19 patients is available, it is feasible to assume a worse prognosis for patients with LGE than those without LGE or healthy controls, considering already published data.Such worsening could involve a higher incidence of heart failure or an increased chance of sudden cardiac death, possibly reducing the life expectancy in these patients.

Conclusions
More than two years after the outbreak, there is still a lack of consensus about how CMR-derived indicators may signal cardiac involvement in patients who recovered from COVID-19.However, most of the selected articles in this review report some extent of myocardial injury in these patients, regardless of conflicting or ambiguous data.With hundreds of millions of cases to date, and a growing number of cases, myocardial involvement could present a threat and heavy burden for healthcare systems worldwide.
Extensive, comprehensive multicenter prospective studies are still needed to understand how myocardial involvement affects patients who recovered from COVID-19.With new variants seemingly more contagious, though with a decreased rate of hospitalizations and mortality, further studies must be performed, ranging from asymptomatic to severe cases.In addition, considering that signs of myocardial injury are already linked with poor prognosis in different cardiac diagnoses, follow-up studies of these patients, especially those with LGE, could aid the early identification of persistent or developing cardiac pathologies.

Table 4 . Reported LGE data.
[25] patients who recovered from COVID-19 had LGE with non-specific distribution and appearance.Further data regarding CMR findings show significant differences in LVEF, RVEF, and LVSV between recovered patients with LGE, without LGE, and controls.Breitbart et al.[25], 2021 7 (12.5%)post-COVID-19 patients with no history of previous heart disease had LGE.