Myocardial Injury Pattern at MRI in COVID-19 Vaccine-Associated Myocarditis.

Background There are limited data on the pattern and severity of myocardial injury in patients with COVID-19 vaccination-associated myocarditis. Purpose To describe myocardial injury following COVID-19 vaccination and to compare these findings to other causes of myocarditis. Materials and Methods In this retrospective cohort study, consecutive adult patients with myocarditis with at least one T1-based and at least one T2-based abnormality at cardiac MRI performed at a tertiary referral hospital from December 2019 to November 2021 were included. Patients were classified into one of three groups: myocarditis following COVID-19 vaccination, myocarditis following COVID-19 illness, and other myocarditis not associated with COVID-19 vaccination or illness. Results Of the 92 included patients, 21 (23%) had myocarditis following COVID-19 vaccination (mean age, 31 years ± 14 [SD]; 17 men; messenger RNA-1273 in 12 [57%] and BNT162b2 in nine [43%]). Ten of 92 (11%) patients had myocarditis following COVID-19 illness (mean age, 51 years ± 14; three men) and 61 of 92 (66%) patients had other myocarditis (mean age, 44 years ± 18; 36 men). MRI findings in the 21 patients with vaccine-associated myocarditis included late gadolinium enhancement (LGE) in 17 patients (81%) and left ventricular dysfunction in six (29%). Compared with other causes of myocarditis, patients with vaccine-associated myocarditis had a higher left ventricular ejection fraction and less extensive LGE, even after controlling for age, sex, and time from symptom onset to MRI. The most frequent location of LGE in all groups was subepicardial at the basal inferolateral wall, although septal involvement was less common in vaccine-associated myocarditis. At short-term follow-up (median, 22 days [IQR, 7-48 days]), all patients with vaccine-associated myocarditis were asymptomatic with no adverse events. Conclusion Cardiac MRI demonstrated a similar pattern of myocardial injury in vaccine-associated myocarditis compared with other causes, although abnormalities were less severe, with less frequent septal involvement and no adverse events over the short-term follow-up. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Raman and Neubauer in this issue.

Summary Statement

The pattern of myocardial injury in patients after COVID-19 vaccination at MRI was similar to other causes of myocarditis, but with less severity.

In a retrospective study of 92 patients with myocarditis, cardiac MRI demonstrated a similar pattern of injury in 21 patients with myocarditis following COVID-19 vaccination compared to other causes, including subepicardial LGE.

Myocardial abnormalities were less severe in patients with vaccine-associated myocarditis (eg, less functional impairment, lower native T1, and less frequent involvement of the septum) compared to other forms of myocarditis.

Introduction

Myocarditis is a non-ischemic inflammatory disease of the myocardium, with diverse causes, clinical patterns, and outcomes (1). Characteristic features are inflammation and myocyte damage, which may be mediated both by direct invasion of the myocardium in the setting of viral infection and by the host's immune response (2). Acute myocarditis is more common in men compared to women, although the incidence is difficult to establish as the clinical presentation is often non-specific and endomyocardial biopsy is not routinely performed (3).

Myocarditis following immunization is a rare event which has received increased attention recently due to reports of myocardial injury in a minority of patients following administration of messenger RNA (mRNA) based COVID-19 vaccines (4,5). As of December 2021, over 4.5 billion people worldwide have received a dose of COVID-19 vaccine (6). Therefore, serious adverse events associated with administration of vaccines targeting COVID-19 are highly relevant to the public, clinicians, and other policy makers, even if the incidence is rare. Importantly, COVID-19 illness can also result in myocardial injury, which is associated with adverse outcomes in hospitalized patients, and should be balanced against the risk of vaccine-related complications (7,8).

Cardiac MRI has an important role in the assessment of acute myocarditis with unparalleled ability for non-invasive characterization of myocardial tissue (9). Several recent case series have described MRI findings in hospitalized patients with myocarditis following COVID-19 vaccination (10-12). However, there are limited data on the extent of myocardial injury in comparison to other causes of myocarditis, particularly in non-hospitalized patients. Understanding the pattern and extent of myocardial injury and its implications will allow for improved care of these patients and may help to address vaccine hesitancy.

The purpose of this study is to determine the pattern and extent of MRI findings in myocarditis associated with COVID-19 vaccination and to compare these findings to other causes of myocarditis.

Methods

Study Design and Participants

This retrospective cohort study was approved by the institutional ethics committee and the requirement for written informed consent was waived. Consecutive adult (≥18 years of age) hospitalized or non-hospitalized patients who were referred to a tertiary hospital network for evaluation of myocarditis by cardiac MRI between December 2019 and November 2021 were identified. Inclusion criteria were fulfillment of 1) clinical presentation and diagnostic testing criteria of the European Society of Cardiology diagnostic criteria for clinically suspected myocarditis (13) and 2) both of the main revised Lake Louise criteria for non-ischemic myocardial inflammation on MRI (at least one T1-based criteria and at least one T2-based criteria, additional details in Appendix E1 (14). Exclusion criteria included MRI performed for follow-up of previously diagnosed myocarditis.

Data on demographic characteristics, vaccine administration, medications, blood test results, ECG parameters, and clinical outcomes were extracted from the electronic patient record. Patients were classified into one of three groups: COVID-19 vaccine associated myocarditis (symptom onset within 14 days of vaccine administration with no other cause for myocarditis identified), COVID-19 illness associated myocarditis (symptom onset within 14 days of confirmed severe acute respiratory syndrome coronavirus 2 [SARS-COV-2] infection based on reverse-transcriptase-polymerase chain reaction assays of nasopharyngeal swabs with no other cause for myocarditis identified), and other myocarditis (all other patients meeting inclusion criteria without temporally associated COVID-19 vaccine administration or known COVID-19 illness) (15,16). Adverse cardiac events were evaluated at short-term follow-up, including death, arrhythmia (defined as sustained atrial or ventricular arrhythmia lasting at least 30 seconds), and heart failure hospitalization. Clinically available blood tests including high sensitivity cardiac troponin I (hsTnI), B-type natriuretic peptide (BNP), and C-reactive protein (CRP) were collected.

MRI Technique

Cardiac MRI studies were performed using 1.5 or 3 Tesla scanners (Magnetom AVANTOfit/SKYRAfit, Siemens Healthineers, Siemens, Germany) with commercially available cardiac surface coils. The MRI protocol included long-axis and a stack of short-axis balanced cine steady state-free precession (bSSFP) slices (slice thickness 8 mm and 2 mm inter-slice gap) and a stack of black-blood T2-weighted spectral attenuated inversion-recovery (SPAIR) images at matching short-axis locations. Late gadolinium enhanced (LGE) images were acquired using a 2D phase sensitive inversion recovery technique starting 12 minutes after administration of intravenous contrast (0.15 mmol/kg body weight of gadobutrol, Bayer Healthcare, Berlin, Germany) (17). A single mid-ventricular short-axis T1 and T2 mapping slice was acquired using a modified Look-Locker Inversion Recovery (MOLLI) technique for native T1 mapping (5(3)3 inversion grouping) (18) and a matching T2 map using a T2-prep technique with read-out varying with external field-strength (bSSFP at 1.5T and Fast Low Angle Shot [FLASH] at 3T) (19). Pixel based T1 and T2 maps were automatically generated on the scanner with application of inline motion correction algorithms.

MRI Analysis

MRI studies were analyzed independently by two experienced fellowship trained observers [author initials blinded for review] who were blinded to all clinical information using commercially available tools (Circle cmr42; Circle Cardiovascular Imaging, Calgary, Canada). Ventricular volumes, function, and mass were measured using semi-automated contour detection with manual correction if required as per established standards (20). Global longitudinal, circumferential, and radial strain (GLS [3 long-axis views], GRS, and GCS [entire short-axis stack], respectively) were calculated from bSSFP images using feature tracking strain analysis. Presence of LGE and regional T2-weighted hyperintensity were evaluated visually (present or not) globally and according to the AHA 17-segment model (21). For assessment of LGE, the predominant pattern was classified as subendocardial, mid-wall, subepicardial, or transmural. LGE was quantified using a signal-intensity threshold of 4 standard deviations (SD) above visually normal reference myocardium, expressed in grams and as a percentage of left ventricular mass. Source T1 and T2 mapping images were examined for artifacts and any segments with artifact were excluded from analysis. Septal T1 and T2 relaxation times were assessed by manually drawing a region of interest at the mid-interventricular septum avoiding the right ventricular insertion points and blood pool. Maximum T1 and T2 values were also measured by manually drawing a region of interest in areas of visually maximum myocardial values based on the color map, with a minimum region of interest size of 0.5 cm2. As per current guidelines, abnormal maximum T1 and T2 values were defined as 2 SD above the mean of sequence specific local reference values (high T2 defined as >52 ms at 1.5T and >45 ms at 3T; high T1 defined as >1067 ms at 1.5T and >1289 ms at 3T) (22). To facilitate combined analysis of multi-scanner data, T1 and T2 values were converted to a z-score using scanner-specific local reference values (patient value − mean of reference range)/(SD of reference range) (23). In this case, z-scores provide an assessment of how many SD each patient's T1 or T2 value is above or below the mean for the normal range for each scanner.

Statistical analysis

Categorical data are presented as counts (percentages) and continuous variables as means (standard deviation) and medians (interquartile ranges [IQRs]). All continuous data were tested for normal distribution using the Shapiro-Wilk test. Comparisons between groups were conducted using one-way analysis of variance for continuous variables with normal distribution and Kruskal-Wallis test for continuous variables with non-normal distribution, with post hoc tests for significance between groups using Bonferroni correction. Fisher's exact test was used to compare categorical variables. We performed sensitivity analyses restricting the vaccine associated group to those without a prior history of COVID-19 infection and restricting the other myocarditis group to those with a non-COVID-19 viral/post-infectious cause of myocarditis. Spearman correlation analysis was used to evaluate associations between continuous variables. Linear regression was used to evaluate the relationship between continuous MRI parameters and patient group controlling for patient age, sex, and length of time from symptoms to MRI. All tests were 2-tailed, and P values less than .05 were considered statistically significant. Analysis was performed using STATA software v14.1 (StataCorp, College Station, Texas) and data were visualized with GraphPad Prism 9.0.2 (GraphPad Software, Inc., La Jolla, CA, USA).

Results

Patient Characteristics

Ninety-six patients were evaluated for eligibility. Four patients were excluded as MRI was performed for follow-up of previously diagnosed myocarditis (Figure 1) leaving 92 patients (mean age, 41±18 years, 61% men), Table 1. Twenty-one patients (23%) had COVID-19 vaccine associated myocarditis, of which 17 (81%) were male and mean (SD) age was 31 (14) years; 10 (11%) had myocardial injury following COVID-19 illness and 61 (66%) had myocarditis not temporally associated with either COVID-19 vaccination or COVID-19 illness (non-COVID-19 viral/post-infectious in 19 [31%], autoimmune in 8 [13%], drug related in 6 [10%], hyper-eosinophilic in 3 [5%], and other/unknown in 25 [41%]). Patients with vaccine associated myocarditis were younger and more frequently male compared to the other groups. The median [IQR] interval between symptom onset and MRI was 11 [4-29] days.

Figure 1.

Flow chart detailing patient selection.

Table 1.

Patient Characteristics and Blood Test Results

COVID-19 Vaccine Associated Myocarditis

Among patients with vaccine associated myocarditis, symptom onset followed administration of mRNA-1273 (Moderna) in 12 (57%) and BNT162b2 (Pfizer-BioNTech) in 9 (43%) and occurred after administration of the second vaccine dose of a 2-dose series in 17 patients (81%). Those who had received two doses had a median [IQR] interval between first and second doses of 33 [25-41] days. Two patients had a remote history of COVID-19 illness, both of whom had mild disease severity and recovered at home with interval between COVID-19 diagnosis and vaccine administration of 111 days and 155 days. Chest pain occurred in all 21 patients and started at median [IQR] of 3 [1-7] days after vaccination and lasted 1 to 6 days. Fourteen patients (67%) were admitted to hospital with median [IQR] length of stay of 3 [2-5] days. No patients were admitted to the intensive care unit. Ten patients were treated with colchicine (48%), seven with aspirin (35%), four with ibuprofen (20%), and one with steroids (5%). Troponin levels were elevated in all patients admitted to hospital (>26 pg/mL) and substantially decreased by the time of discharge (median [IQR] 2723 [1500-5772] pg/mL vs 49 [0-205] pg/mL; P =0.001).

Cardiac MRI of patients with Vaccine Associated Myocarditis

MRI characteristics are provided in Table 2. Abnormal MRI findings among patients with myocarditis following COVID-19 vaccination included LGE in 17 (81%), high T1 in 14 (67%), high T2 in 16 (76%), hyperintense signal on T2-weighted imaging in 15 (79%), and systolic left ventricular dysfunction (LVEF <55%) in 6 (29%), Figure 2. In all patients, at least one T2-based abnormality co-localized within the same myocardial segment as a T1-based abnormality, including LGE. None of the MRI parameters differed significantly between vaccine types.

Table 2.

Cardiac MRI Findings

Figure 2.

COVID-19 vaccine associated myocarditis. Short-axis 1.5T MRI images and ECG findings of a 19-year-old man with myopericarditis who presented with chest pain 3 days following the second dose of an mRNA COVID-19 vaccine (mRNA-1273). Cardiac MRI performed 2 days after symptom onset demonstrates mid wall to subepicardial late gadolinium enhancement (LGE) at the basal to mid inferior lateral wall with adjacent pericardial enhancement (A, red arrow), corresponding hyperintensity on T2-weighted imaging (B, orange arrows), abnormal high native T1 (C, 1095 ms, maximum region of interest), and abnormal high native T2 (D, 57 ms, maximum region of interest). The ECG demonstrates diffuse concave upward ST segment elevation except in leads aVR and V1, upright T waves in the leads with ST segment elevation, and PR depression, consistent with pericarditis (E). Peak high sensitivity troponin-I was 5772 pg/mL. He was admitted to hospital and was discharged after two days following complete resolution of his symptoms. At short-interval follow-up he was asymptomatic with normal troponin levels.

Peak hsTnI correlated significantly with maximum native T2 z-score (r=0.50; P=.040), LVEF (r=-0.58; P=.015), GCS (r = 0.66; P = .005), and GRS (r = -0.59; P = .013), but not with maximum native T1 z-score, LGE extent, or GLS.

Comparison to COVID-19 and Other Causes of Myocarditis

Compared with patients with other causes of myocarditis, patients with vaccine associated myocarditis had significantly higher LVEF and RVEF, less impaired GLS, GCS and GRS, lower native T1, and less extensive LGE, Figure 3, Table 2. Differences in LVEF, GLS, GCS, GRS, native T1, and LGE extent remained significant even after controlling for patient age, sex, and interval between symptom onset and imaging (Table 3). Compared to patients with COVID-19 illness, patients with vaccine associated myocarditis had higher LVEF, less regional wall motion abnormalities, and lower native T1. Differences in LVEF remained significant even in the multivariable model. In all three patient groups, the most frequent pattern of LGE was subepicardial and the most frequent myocardial segment involved was the basal inferolateral wall, Figure 4. However, patients with COVID-19 illness and other myocarditis had a higher prevalence of abnormalities involving the basal to mid anterior and inferior septum, while patients with vaccine associated myocarditis rarely had abnormalities involving the anterior wall or septum. There were no significant differences in blood biomarkers or ECG parameters between groups.

Figure 3.

Scatterplots of (A) LVEF, (B) LGE, (C) Native T1, and (D) Native T2 by Group. Graphs for MRI parameters depict individual patient data points with error bars displayed as median and interquartile range. There were significant differences in maximum native T1 z-score, maximum native T2 z-score, and LGE extent (as a percentage of left ventricular mass) between patients with vaccine associated myocarditis (vaccine) and patients with other myocarditis (other), denoted with ***. All other comparisons between patients with vaccine associated myocarditis and patients with COVID-19 illness (COVID-19) or other myocarditis were not significant, denoted with ns. LGE, late gadolinium enhancement; LVEF, left ventricular ejection fraction.

Table 3.

Univariable and Multivariable Linear Regression Parameters with Vaccine Associated Myocarditis as the Reference Group

Figure 4.

Segmental Distribution of MRI abnormalities. Color shaded bulls-eye plots represent the percentage of patients in each patient group with late gadolinium enhancement (LGE) and/or hyperintensity on T2-weighted imaging for each myocardial segment according to a standardized 17-segment model. COVID-19 vaccine, patients with vaccine associated myocarditis; COVID-19 illness, patients recovered from COVID-19; other myocarditis, patients with other causes of myocarditis.

Sensitivity Analysis

Conclusions of our primary analyses were unchanged upon removing the 2 patients with vaccine associated myocarditis with a prior history of COVID-19 illness (Table E1) and when the other myocarditis group was restricted to patients with non-COVID-19 viral/post-infectious myocarditis (Table E2).

Table E1.

Patients Vaccine Associated Myocarditis with Prior COVID-19 Infection (n=2) Removed from the Analysis

Table E2.

Other Myocarditis Group Restricted to Those with Non-COVID-19 Viral Infectious Cause of Myocarditis

Follow-up

All patients with vaccine associated myocarditis had short-term clinical follow-up with median [IQR] follow-up duration of 22 [7-49] days. At follow-up, all 21 patients (100%) were asymptomatic, 8 (38%) had normal troponin levels, and 9 (43%) had reduced but still mildly elevated troponin levels (follow-up troponin levels were not available in 4 patients). Of the patients with impaired LVEF on MRI, 4/6 had subsequent transthoracic echocardiography or follow-up MRI which demonstrated normal LVEF in all. No patient with vaccine associated myocarditis had an adverse cardiac event over short-term follow-up.

Among patients with COVID-19 illness and other myocarditis, 7 and 42 had clinical follow-up with median [IQR] follow-up duration of 211 [94-295] days and 195 [87-415] days, respectively. Patients with COVID-19 illness had 3 MACE events (1 hospitalization for heart failure and 2 arrhythmia events [one patient subsequently had an ICD implanted]) while patients with other myocarditis had 5 MACE events (2 hospitalizations for heart failure and 3 arrythmia events [4 patients subsequently had an ICD implanted]). There were no deaths in any group. As expected, the follow-up duration for the other two groups was much longer than for the vaccine group, given the relatively short time interval over which COVID-19 vaccines have been administered.

Discussion

In this retrospective cohort study of 92 patients meeting both clinical and imaging diagnostic criteria for acute myocarditis, we identified 21 patients with myocarditis following COVID-19 vaccine administration who were younger and more frequently male compared to 10 patients who had recovered from COVID-19 illness and 61 patients with other causes of myocarditis. To our knowledge this is the first report of cardiac MRI findings in both hospitalized and non-hospitalized patients with myocarditis following COVID-19 vaccination in comparison to patients with other causes of myocarditis including COVID-19 illness. Abnormal MRI findings among patients with myocarditis following COVID-19 vaccination included LGE in 81%, high T1 in 67%, high T2 76%, and systolic left ventricular dysfunction in 29%. MRI revealed a similar pattern of myocardial injury in patients with myocarditis following COVID-19 vaccination compared to other causes, including subepicardial LGE and edema at the basal inferior lateral wall, although patient demographics differed and abnormalities were less severe. Patients with vaccine associated myocarditis had higher left ventricular ejection fraction and lower native T1 values compared to those with COVID-19 illness and other causes of myocarditis, and demonstrated rapid clinical improvement with no adverse events over short-term follow-up.

Our observations are concordant with case series of hospitalized patients showing that most patients with vaccine associated myocarditis are younger men presenting after the second dose, with frequent presence of LGE and myocardial edema on MRI, and rapid improvement in clinical symptoms on short-term follow-up (10,11,24). Other MRI findings in our cohort included high T1 and T2 mapping values and impaired myocardial strain. T1 and T2 mapping are quantitative tissue characterization techniques that provide complementary information, particularly in the setting of myocardial inflammation. High T2 is specific for increased tissue water and can discriminate between active and healed myocarditis (25). Native T1 is also elevated in the setting of edema, although unlike T2, this change is not specific for acute inflammation and can alternatively reflect fibrosis or infiltration. This might account for the significant correlation of peak troponin with native T2 but not with native T1 in our study.

Unlike prior reports, our findings also demonstrate that myocardial injury is detectable in patients with acute myocarditis not requiring hospital admission and that the severity of MRI abnormalities is milder in general compared to patients with other causes of myocarditis, even after controlling for age, sex and interval between symptom onset and imaging. Patients with myocarditis following COVID-19 illness had lower LVEF, higher prevalence of regional left ventricular dysfunction, and higher native T1 compared to those with vaccine associated myocarditis, although other MRI parameters did not differ significantly. This suggests that the imaging phenotype of patients with COVID-19 related myocardial injury may be intermediate between vaccine associated myocarditis and other causes.

Presentation after the second vaccine dose or after the first dose in the context of prior SARS-CoV-2 infection in most patients indicates that prior exposure is relevant and necessitates continued surveillance for post-vaccination myocarditis particularly following booster doses. Although non-mRNA vaccines were also administered in Canada, all patients with vaccine associated myocarditis in our cohort presented after administration of an mRNA-based COVID-19 vaccine. The mechanisms by which the host's immunologic response to mRNA-based COVID-19 vaccines could lead to myocarditis in a small minority of patients warrants further study, particularly given that other mRNA-based vaccines and therapies are in development (26,27).

Milder MRI abnormalities in patients with vaccine associated myocarditis compared to other causes raises the possibility that this group may have a lower future adverse event rate. Lack of any adverse events in our patients with vaccine associated myocarditis over short-term follow-up is reassuring. However, longer term follow-up is needed, particularly given the association of LGE with adverse cardiac events in non-vaccine associated myocarditis (28,29). Interestingly, one prior study found that patchy and mid-wall, but not subepicardial, patterns of LGE were associated with adverse events in patients with non-vaccine myocarditis (30). Similarly, septal but not lateral LGE location was associated with major adverse cardiac events (30). This requires further investigation given that the majority of patients with vaccine associated myocarditis in our study had a subepicardial pattern of LGE with a predilection for the basal to mid lateral wall and infrequent involvement of the septum, which may be associated with relatively favorable outcomes. LGE usually reflects fibrosis in ischemic and non-ischemic cardiomyopathies; however, in patients with acute myocarditis it often reflects an increased volume of distribution of gadolinium-based contrast agent in the affected region related to myocyte necrosis and edema. In all patients with vaccine associated myocarditis, LGE co-localized with edema in at least one segment, which is associated with better prognosis compared to isolated LGE without T2-hyperintensity, and confers the possibility of recovery as edema improves with time (31).

Our study has limitations including a modest sample size and short-interval follow-up among patients with myocarditis following vaccination. More than one MRI scanner was used for imaging which impacts quantitative parametric mapping. To address this, we interpreted mapping values in the context of scanner specific local reference ranges and calculated z-scores for T1 and T2 values. The timing of MRI after symptom onset varied which could impact detection of myocardial edema, as MRI markers of edema typically demonstrate rapid improvement during the first few weeks after symptom onset (32). There were also significant differences in patient age and sex between groups. Although we adjusted for age, sex and timing of imaging in our analysis, it is possible that this did not fully account for differences between groups. Only mid-ventricular T1 and T2 mapping slices were examined, which could underestimate maximum T1 and T2 values if regional disease was only present in other areas of myocardium. However, LGE and T2-weighted imaging were performed with coverage of the entire myocardium from base to apex. This may account for the overall slightly higher prevalence of abnormalities on LGE and T2-weighted imaging compared to T1 and T2 mapping, respectively. As this was not a population-based study, we could not calculate the incidence of vaccine associated myocarditis. There is no standardized definition of vaccine associated myocarditis or COVID-19 related myocardial injury in the literature to date with respect to the timing of symptom onset after vaccine administration or COVID-19 diagnosis. For consistency, we used the same 14-day interval between vaccination and COVID-19 diagnosis to symptom onset to define both groups. Finally, histologic confirmation of myocarditis was not available as endomyocardial biopsy is not frequently performed at our center unless there is clinical evidence that the results will have a meaningful effect on therapeutic decisions (33). Our findings should be confirmed in future large, multi-center studies with longer term follow-up.

In conclusion, our study demonstrates that the pattern of MRI abnormalities in vaccine associated myocarditis is similar to other causes although patient demographics differ and MRI findings tend to be less severe. Overall, our study findings are generally consistent with an imaging phenotype with good prognosis; however, further studies are needed to examine the long-term effects of mRNA-based COVID-19 vaccination on the heart, determine the risk associated with booster doses, and inform recommendations for vaccination in patients with a history of myocarditis.