Imaging Findings in Pediatric COVID-19: A Review of Current Literature

The cardiovascular system is one of the organ systems that appears to be particularly affected by SARS-CoV-2. The direct mechanism of cardiac involvement is unclear, but there have been several proposed mechanisms. Some involve a cytokine storm phenomenon that leads to subsequent cardiac involvement similar to Kawasaki disease [14, 15]. Another theory posits that spike protein, via the ACE2-receptor binding domain found in alveolar lung tissue and myocardial tissue [16], binds to cardiomyocytes in a similar fashion as has been seen for SARS-CoV-1 [17]. Oudit and colleagues demonstrated myocardial susceptibility when they demonstrated 35% of heart biopsy specimens collected from the 2009 SARS-CoV-1 outbreak carried the SARS-CoV-1 genome [17]. Recent studies have produced similar results, with 62% of cardiac biopsy specimens documenting SARS-CoV-2 viral load and nearly half of these having what was classified as high viral copy numbers [18]. In the study by Lindner et al., myocardial involvement was found without markers of fulminant myocarditis [18]. This feature may speak to the cardiotropic effects of SARS-CoV-2 separate from a fulminant myocarditis-type effect that has been previously described [19]. Baily et al. describe cardiomyocyte infection in an engineered heart tissue model, resulting in intracellular cytokine production, sarcomere disassembly, contractile deficits, and cell death [20]. Further work will be needed to delineate the exact mechanism, and whether different mechanisms are correlated to different disease phenotypes within COVID-19. The verdict is still out on whether ACE-2 is a feature of proposed cardiotropic effect of SARS-CoV-2, with some studies supporting direct involvement of ACE-2 [21, 22] and some via a secondary down-stream change [16]. Additionally, future studies will need to focus on factors such as patient-specific genomics and how these may affect susceptibility to different disease states within COVID-19.

The main cardiac imaging modalities utilized are transthoracic echocardiography (TTE) and cardiac MRI (CMR). TTE offers the advantages of being widely available, portable, requiring a shorter scan time, and providing a good evaluation of myocardial and valvar function. Several studies have noted variable findings by TTE in COVID-19 patients. Stobe et al. demonstrated that a cohort of COVID-19 patients demonstrated abnormal left ventricular echocardiographic strain in the basal segments [23]. Seventy-one percent of their cohort demonstrated abnormal strain parameters, despite normal left ventricular ejection fraction. However, while TTE is often the first-line cardiac imaging modality for a variety of disease processes, its main limitation is its inability to assess for changes in the myocardial tissue.

The evidence of the cardiotropic nature of SARS-CoV-2 requires accurate assessment of myocardial involvement in COVID-19 patients and this is where CMR plays a key role. Evaluation of the myocardium requires great equipoise between the invasiveness of testing and diagnostic accuracy, particularly in the pediatric patient. While endomyocardial biopsy is considered the reference standard for myocardial tissue characterization, innovations in CMR (including new sequences and rapid acquisition techniques) coupled with experience and expertise have brought non-invasive evaluation to the forefront. The recent AHA Scientific Statement on Pediatric Myocarditis is a prime example of the prominent role that CMR plays in evaluating the myocardium, elevating CMR findings close to the reference standard of endomyocardial biopsy [24]. These recommendations mirror the current clinical practice of shifting away from endomyocardial biopsy to less-invasive diagnostics. This recommendation is supported by the myriad of literature demonstrating accuracy of tissue characterization by CMR. Ferreira and colleagues provided an update to the previous standard CMR guidelines known as the Lake Louise Criteria (LLC), describing tissue characteristics of patients with nonischemic myocardial inflammation [24]. Briefly, myocardial inflammation alters the T1 and T2 relaxation times of the myocardium measured by CMR imaging compared to normal reference T1 and T2 values for the myocardium (Fig. 1). When inflammation leads to myocyte injury, this feature manifests as late gadolinium enhancement (LGE), one of the keystones for myocardial tissue characterization by CMR [25]. Late gadolinium enhancement is the reference standard for myocardial viability assessment and is one of the most widely utilized CMR techniques. These features combined are indicative of myocardial edema and necrosis [26].

Given this and other studies highlighting the benefit of non-invasive imaging, current guidelines state that CMR carries the same class 1 indication as endomyocardial biopsy for the diagnosis of myocarditis [24]. Several studies have utilized CMR for myocardial evaluation in the COVID-19 patient and demonstrated important findings within the myocardium after SARS-CoV-2 infection. We review some of the important evidence of cardiac involvement of patients suffering from COVID-19 and seek to highlight the different non-invasive imaging findings within different subsets of this cardiac involvement. We will also call to attention areas in the literature that require further studies, as we all continue to rapidly learn different features of this new disease. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Several studies have highlighted that patients presenting with elevated serum cardiac biomarkers are more likely to have findings consistent with myocarditis as defined as myocardial edema or other features of positive T2 criterion, which is the most common finding in the COVID-19 myocarditis patient, in addition to at least one T1-based criterion [33, 34]. Chen and colleagues describe a 25-patient cohort of adults who had CMR performed in the acute phase (defined in their study as 3–8 days after diagnosis) of COVID-19 infection. Patients with at least one marker of cardiac involvement had significantly elevated T1 relaxation time, extracellular volume, T2 mapping, and LGE in addition to worse global longitudinal strain [34]. Importantly, these findings were significantly increased compared to healthy controls, irrespective of troponin values. The average age in this study was 23 years old, highlighting the lack of protection for the young adult patient against cardiac involvement.

Cases of cardiac involvement harbor a poor prognostic factor in the adult population [27, 28, 31, 35‐37], and the prevalence of cardiac involvement has been on the rise. The odds of myocarditis diagnosis has been found to be 16 times greater for those with COVID-19 than those without COVID [38]. In a large study of hospitalized COVID-19 adults, those with cardiac injury had a 59% incidence of acute respiratory distress syndrome (ARDS) and a 51% mortality rate, compared to 15% and 5%, respectively, in those without cardiac injury [39].

Due to the concern for SARS-CoV-2 having a predilection for affecting the myocardium and the risk of exercise precipitating arrhythmias [40], a subset of COVID-19 patients that has encountered specific controversy is the asymptomatic or minimally symptomatic competitive athlete recovering from COVID-19. There is great uncertainty surrounding how to screen these patients for disease, how to determine cardiac involvement, and how to counsel these athletes regarding return to sporting activity. This involvement has extended all the way to the professional level [41]. Expert recommendations, as early as fall 2020, have recommended a tiered approach with clinical symptoms, serum biomarkers, and electrocardiogram (ECG)/echocardiographic assessment to guide which patients warrant further imaging via CMR [40‐42].

Daniels and colleagues report on a large cohort of collegiate athletes from the Big Ten Conference undergoing screening after SARS-CoV-2 infection. While there was some variation in screening strategy, a large number of athletes underwent primary CMR screening. They found COVID-19 myocarditis, as defined by LLC on CMR, in 2.3% of athletes amongst the Big Ten COVID-19 cardiac registry [43]. Twenty-eight of the 37 myocarditis cases were asymptomatic and the diagnosis was made on CMR findings alone, where all other imaging modalities were normal, highlighting the prevalence of subclinical myocarditis in this disease and the importance of CMR. Follow-up CMR on these patients showed that all had resolution of markers of myocardial edema, but that 60% had continued myocardial scar. The authors describe several considerations regarding these findings, and importantly note the risk of sudden cardiac death, even in asymptomatic athletes with myocarditis, as was described to occur in nearly half of viral myocarditis cases of sudden death in young patients [44]. The occult nature of CMR findings may provide further insight into a possible cardiotropic nature of SARS-CoV-2, thus allowing cardiac involvement without overt involvement of other organs. With this in mind, the risk of sudden death during exercise must be considered when determining return-to-play recommendations for the COVID-19 athlete. Malek and colleagues similarly noted CMR abnormalities at follow-up MRI in 19% of elite athletes despite mild/asymptomatic COVID-19 infection in the vast majority of participants [45]. However, rates of cardiac involvement in other studies have not been as high as other studies [46]. As others have noted [47], the clinical significance of these CMR findings are yet to be identified for the competitive athlete, and further studies will be required to delineate restrictions and return-to-play given the importance prevalence of myocarditis in the competitive athlete after COVID-19. There is a paucity of data showing that such features as isolated myocardial edema affect long-term prognosis in other cases of myocarditis [45]. The possibility of ongoing inflammation and a potential nidus for dysrhythmia should be entertained, but this will need to be balanced with the notable likelihood of false-positive findings on CMR. Of the 97 deaths due to viral myocarditis described in a study by Harris et al., 58 were physically active at or near time of death [48], bringing to light the importance of safe return to play. Nearly half of these sudden unexpected deaths in myocarditis patients were precluded by a viral prodrome, which may assist in restricting athletes that continue to be symptomatic following infection and/or the “Long Covid” subgroup of COVID-19 patients [49]. These findings are summarized in Table 1.

Despite presenting with significant dysfunction, normalization of left ventricular ejection fraction (LVEF) was achieved in nearly all patients following recovery from MIS-C. Findings from Belhadjer and colleagues described a 35-patient cohort of patients, median age 10 years, diagnosed with MIS-C. All 35 patients in this cohort presented with fever and had a LVEF of < 50%, with 28% having an LVEF of < 30%. Eighty percent of these patients required inotropic support and a striking 28% required extracorporeal membrane oxygenation (ECMO). A total of four of 35 showed segmental wall hypokinesis in addition to the left ventricular dysfunction. Recovery in LVEF was seen in 71% of the patients and was achieved at a median of 2 days, which is a strikingly fast recovery rate for patients presenting with such severe features. Regional wall motion abnormalities have been described in 7–10% of MIS-C patients, and 35% of patients have been found to have arrhythmias during hospitalization [53]. Several studies evaluating echocardiographic strain demonstrated residual abnormalities in these patients, particularly those who presented with more severe illness, speaking further to residual occult disease [60, 61]. Sanil et al. found that those who had worse left ventricular longitudinal strain (LVGLS) on admission had higher peak troponin elevation and less improvement in LVGLS by 10 weeks of follow-up.

Right heart involvement has yet to clearly been demonstrated in MIS-C, as the majority of the pathologic findings appear isolated to the left ventricle. When involved, changes in right ventricular parameters such as abnormal echocardiographic strain has been predictive of myocardial injury [60] and has been shown to be a predictor of mortality in adult patients [62]. Further pediatric echocardiography studies regarding this topic are on-going.

The presence of coronary artery involvement in the form of coronary ectasia or aneurysm formation is another prominent imaging finding of MIS-C. A large cross-sectional study of MIS-C patients in the USA found a coronary artery abnormality prevalence of 16.5% [63]. Although less frequent, giant coronary aneurysms have been reported with MIS-C and based on literature from the Kawasaki disease population, are the most likely to have long-term sequelae [64]. Figure 1 describes aneurysmal changes in two different MIS-C patients.

CMR assessment of MIS-C has followed shortly behind the clinical and echocardiographic assessment of the disease, as we continue to learn more about the subacute and longer-term course of these patients. Early reports from the United Kingdom described varying findings in CMR parameters when assessing mean T1 and T2 mapping. A large multicenter MIS-C registry from Europe included 42 pediatric patients with CMR and showed myocardial edema via T2 mapping to be present in 33% and LGE in 14% of patient during the acute hospitalization [53]. A study by Bermejo and colleagues performed at a mean of 27 days after onset of symptoms showed no increased mean T1 and T2 mapping, but 2 of 4 did show LGE [65]. Complicating interpretation was that all but one of these 20 patients described by Bermejo and colleagues had normal biventricular systolic function as assessed by CMR, despite half of the patients having reduced LVEF by transthoracic echocardiography. Upon further review, two of the 20 were found to have segmental T2 mapping abnormalities. These areas were associated with elevated C-reactive protein (CRP) at initial presentation. A smaller study of four patients in the acute period of MIS-C found that 75% of patients had elevated mean T1 and/or T2 mapping, without evidence of LGE in the acute phase [66]. Similar to the previously noted TTE studies, abnormal strain imaging by CMR has also been reported in MIS-C [67]. The summary of most findings, highlighted by these and others, seem to show acute findings consistent with myocarditis on CMR that resolves with clinical improvement (Fig. 2). These findings are summarized in Table 1. Future study into the long-term recovery of the myocardium and subsequent sequelae is needed.

Previously noted EG and echocardiographic abnormalities from acute infection may persist for some time, however to date, no studies have identified overt heart failure persisting in COVID-19. Some studies have shown both ischemic and non-ischemic changes on CMR for patients recovered from severe COVID-19 infection [32], highlighting the importance in considering cardiac ischemia in the recovered COVID-19 patient with new chest pain or other cardiac symptomatology. Puntmann and colleagues showed that 78% of patients recovered from COVID illness had positive CMR findings at an average of 71 days following symptom onset (Table 1). Similar findings have been shown by Kotecha and colleagues [32]. A large majority of these patients had elevated troponin, possibly indicating a disease subset more likely to show cardiac involvement at follow-up assessment. Importantly, only one-third of the patients in this study had illness requiring hospitalization, showing that disease severity and cardiac involvement are not exclusive. In a study by Huang et al., of 26 patients with moderate-to-severe COVID infection who reported cardiac symptoms during recovery, 58% had abnormal CMR findings at an average follow-up time of 47 days after cardiac symptom onset [76]. None of these patients had prior known myocarditis during their COVID illness. There was no difference found between CMR-positive and CMR-negative groups and different cardiac symptoms, suggesting that symptomatology did not play a factor in predicting CMR findings. Also, no difference was seen in LV function and volumetric assessments, further challenging mechanistic interpretation. The high prevalence of CMR positive findings in this cohort of moderate-to-severe COVID, at an average of 47 days, suggests some type of underlying cardiac involvement and may carry a higher risk in those with residual cardiac symptoms during recovery. Several studies have similarly shown that cardiac symptoms are less likely following recovery of mild COVID, but still occur in 10–20% of cases [74] and cardiac symptoms and/or biomarkers seem to portend higher likelihood of CMR findings [77]. Further studies, as discussed previously, are likely to center around mechanisms surrounding cardiotropic effects of ACE2 receptor binding domain of spike protein versus the inflammatory storm induced in some patients suffering COVID-19 infection [17, 37].

Challenging this inclination are studies such as those performed by Joy and colleagues [29]. They found that when comparing mild seropositive infections 6 months following onset to seronegative controls, CMR abnormalities were no more likely in the mild COVID infections compared to healthy controls. It is important to note that this cohort included mild or asymptomatic COVID infections found in healthcare workers, which may limit applicability. Studies such as these lead to the suspicion that mild COVID infection has reduced cardiac pathology 6 + months following infection, but this cannot yet be applied to more severe infection or at a shorter follow-up window.