The pandemic of the coronavirus 2019 disease (COVID-19) is currently spreading across the globe and has had a death toll surpassing 2,000,000 individuals.1 The most common presenting symptoms in patients are fever, cough, myalgia, and fatigue. Typical findings in computed tomography (CT) include bilateral peripheral ground-glass opacities, intralobular and interlobular thickening, crazy-paving pattern, and consolidations.2
There is increasing evidence that COVID-19 affects various organ systems besides the respiratory system, including the heart, with multiorgan involvement playing an important role in disease prognosis.
High-sensitivity troponin T (hsTnT) is a validated serological marker for myocardial injury and routinely used for the early diagnosis of myocardial infarction. However, hsTnT is also elevated in other cardiac diseases including myocarditis and noncardiac disease such as chronic renal insufficiency.3
An observational study from Wuhan, China, found hsTnT to be an independent risk factor with a mortality of about 50% in patients with COVID-19, if hsTnT was elevated at symptom onset and at admission,4 suggesting a cardiac involvement in COVID-19. A potential biological link for this is the ability of the underlying severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to bind the angiotensin-converting enzyme 2 (ACE2) receptor, which is also expressed in cardiomyocytes.5 Another possible mechanism is a systemic inflammatory response with activation of interleukins and interferons leading to a cytokine storm.6 Finally, a SARS-CoV2-induced hypercoagulatory state could lead to thrombosis and ischemia.7 Nevertheless, few advances have been made in determining the mechanism of cardiac injury.5 COVID-19 has so far only been detected in the heart in a few cases.8,9 Few studies report nonischemic myocardial injury (NIMI) in COVID-19 patients and individuals recovered from COVID-19.10
Although endomyocardial biopsy (EMB) is considered the gold standard in the diagnosis of myocarditis, cardiac magnetic resonance imaging (CMRI) is the most accurate noninvasive method for its detection.3,11 With a combination of different imaging sequences, it can reach an estimated area-under-the-curve of 96% for the diagnosis of acute myocarditis in non-COVID-19 cohorts.11
The main diagnostic targets of myocarditis in CMRI are myocardial edema (ME), hyperemia with increased capillary leak resulting in increased vascular and extravascular space, and focal necrosis, fibrosis, and scarring. These alterations can be detected by CMRI using a combination of T2-weighted imaging, T1-based imaging with early and late gadolinium enhancement (LGE), and novel T1 and T2 mapping techniques, according to the updated Lake Louise Criteria (LLC).11 We aimed to compare CMRI markers in COVID-19 patients to CMRI alterations in non-COVID-19 patients with clinically suspected myocarditis for examinations performed during the same time period.
MATERIALS AND METHODS
This single-center prospective cohort study was carried out at the BLINDED Hospital between March 31st and May 5th. The study followed the principles of the Declaration of Helsinki. Patients are part of the COVID-19 Registry of the BLINDED Hospital (CORKUM, WHO trial id DRKS00021225). Patient data were anonymized for analysis and the study was approved by the local ethics committee (No: 20-245).
We selected study patients from a consecutive cohort of inpatients who were referred to CMRI for clinical indications due to suspected myocardial injury and polymerase chain reaction (PCR)-confirmed infection. Suspected myocardial injury was based on the following criteria: indeterminate alterations in electrocardiogram, clinical suspicion of myocarditis, or unclear elevation of serum hsTnT in the absence of relevant coronary artery disease (CAD) or previous cardiac afflictions. Written informed consent was required for study inclusion.
Exclusion criteria were repeated negative PCR for SARS-CoV-2, normal hsTnT values at the time of CMRI, known allergy to gadolinium, non-MR compatible foreign material, renal insufficiency (glomerular filtration rate <30 mL/min per 1.73 m2), pregnancy, inability to perform breath holds during the examination or to lie in a supine position due to shortness of breath or coughing, claustrophobia, and cardiorespiratory instability.
As a control group, we retrospectively selected 18 consecutive patients, starting from the date of the first CMRI of a patient with COVID-19 due to suspected myocarditis and who did not suffer from any other cardiac condition (eg, dilated cardiomyopathy, CAD). All patients in the control group with suspected myocarditis had received previous coronary CTA or diagnostic catheterization to exclude relevant CAD.
Magnetic Resonance Imaging
All patients were examined in the supine position with a whole-body 1.5 Tesla MR scanner (MAGNETOM Aera, Siemens Healthineers) using an 18-channel phased-array surface coil. For the acquisition, ECG and diaphragm triggers were used. Cardiac function was analyzed using ECG-gated imaging with the balanced steady state free precession (bSSFP) sequence as a stack in the short axis covering the whole heart in 8 to 10 slices and 1 slice in 2-chamber, 3-chamber, and 4-chamber views (SL 8 mm, TE 1.41ms, TR 30.6 ms). Dark blood fat-saturated spectral attenuated inversion recovery (SPAIR) sequences (T2wfs) in the short axis (3 slices), and the 2-chamber and 4-chamber views were acquired to visualize edema (SL 8 mm, TE 87 ms, TR 1700 ms). T2 mapping was performed using T2-prepared single-shot SSFP sequences with prep pulses at 0, 25, and 55 ms (SL 8 mm, TE 1.060, TR 193.370). T1 mapping was performed before contrast agent administration (T1) and immediately before scar imaging using modified look locker (MOLLI) sequences with a flip angle of 35 degrees and 8 ascending times to inversion (TI) ranging between 100 and 4200 ms (SL 8 mm, TE 1.080 ms, TR 277.320 ms). LGE was visualized with inversion recovery gradient echo pulse sequences ~10 minutes after the intravenous administration of 0.15 mmol/kg bw Gadobutrol (Bayer Vital, Leverkusen, Germany) in a 10-slice short axis stack, and 2-chamber, 3-chamber, and 4-chamber views reconstructing both phase-sensitive and magnitude images (SL 8 mm, TE 3.48 ms, TR 641 ms). Optimal TI for scar imaging was performed via Look-Locker TI-scout. An example for the mapping sequences and LGE is shown in Figure 1.
Image analysis was carried out using commercially available software (cvi42 v.5.12.0; Circle Cardiovascular Imaging Inc.). Functional imaging was analyzed by semiautomatic delineation of the left ventricular (LV) endocardial and epicardial contour in the short-axis stack to assess LV ejection fraction (LVEF), body surface area indexed LV end-diastolic volume (EDVI), LV end-systolic volume (LVESVI), and LV mass (LVMI). In 2 patients with missing short-axis stack, we determined the LV function using the long-axis area-length method.12 T1 and T2 maps were manually reconstructed from the motion-corrected raw data sets. Areas with a correlation of R2<0.7 were excluded from the maps. A scanner-specific and sequence-specific correction factor (1.0365) was used for reconstruction. Values from nonenhanced T1 and T2 maps were compared to scanner-specific normal values established with healthy volunteers. Normal ranges were set at mean ±2.5 SD to compensate for the age difference, resulting in normal ranges of 905 to 1096 ms for T1 and 39 to 54 ms for T2, in line with current recommendations.13 Extracellular volume (ECV) maps were semi-automatically generated from the nonenhanced and postcontrast T1 maps using a hematocrit value from the day of the CMRI.14 For the evaluation of mapping sequences, endocardial and epicardial contours were placed on the edges of the myocardium and an offset of 10% was applied to exclude extramyocardial voxels. To minimize bias, contours were drawn on grayscale maps and map evaluation was performed in different sessions than LGE evaluation. Additional blood pool contours were delineated in non–contrast-enhanced and contrast-enhanced T1-maps to facilitate ECV-map generation. The diagnosis of myocarditis was based on the modified 2018 LLC.11 ME was defined as a quotient of myocardium to skeletal muscle of ≥2.0:1 in T2wfs15 (T2Q) or as an abnormal increase in T2 (T2>54 ms). NIMI was defined either as a visual LGE in typical distribution (subepicardial or patchy midmyocardial), an increase of ECV over 28%,14 or an abnormal increase in T1 (T1>1096 ms). For further analysis, the myocardium was divided into 16 segments (6 basal, 6 midventricular, and 4 apical) according to current American Heart Association (AHA) nomenclature. The mean values for T1, T2, and ECV were determined for each of the 16 AHA-segments. Alterations in over 7 AHA segments were considered global. T2Q was reported as the mean quotient for the whole heart excluding areas with motion artifacts.
Statistical analysis was carried out using IBM SPSS Statistics 24 (IBM Corp.). Normality of distributions was tested using the Shapiro-Wilk Test. Normally distributed variables were compared using the independent t test and non-normally distributed variables were compared using the Mann-Whitney U test. Nominal variables were compared using the χ2 test. Correlation was tested using Pearson Correlation. All values are reported as mean±SD; a P-value of <0.05 was considered statistically significant. The confidence interval was set at 95%. If not otherwise specified, values of the COVID-19 group are reported first.
CMRI was performed in 30 consecutive patients with suspected or confirmed infection with SARS-CoV-2 between 31st March and 31st May 2020. After excluding 6 patients who were examined on clinical suspicion of COVID-19, but had repeated negative PCR testing, 4 patients without elevated hsTnT or ECG abnormalities, and 2 patients who aborted the examination before contrast agent administration, we evaluated 18 CMRI data sets. One examination in the COVID-19 group was performed without mapping sequences and 1 was aborted before LGE. One examination in the control group had faulty non–contrast-enhanced T1 mapping, thus also excluding ECV. All examinations had sufficient data to evaluate NIMI and ME in accordance with LLC.11 In a few patients, some AHA segments could not be evaluated in mapping and LGE sequences due to motion artifacts or partial volume (as reported in Appendix 1, Supplemental Digital Content 1, https://links.lww.com/JTI/A192). Maximum hsTnT and hsTnT at the time of CMRI were not normally distributed. The Shapiro-Wilk test for normal distribution was significant for EF in both groups and for ESVI, mean T1, and mean T2 in the COVID-19 group, but the variables were treated as normally distributed as values showed a linear distribution in the Q-Q plots. All other numerical variables were normally distributed.
Baseline parameters are shown in Table 1. In the COVID-19 group, there were significantly fewer female patients (11% [n=2] vs. 50% [n=9]; P=0.027) and patient age was significantly higher (76.0±4.4 y vs. 50.9±16.2 y; P=0.001). Maximum hsTnT during the admission period (0.111±0.225 ng/mL vs. 0.351±0.496 ng/mL; P=0.129) and at the time of CMRI (0.077±0.196 ng/mL vs. 0.054±0.106 ng/mL; P=0.242) showed no significant difference between both groups. The mean time from symptom onset to CMRI in the COVID-19 group was 22±12.8 days (1 to 41 d). All examined patients with COVID-19 showed symptoms of disease. In the control group, 10 patients (56%) had chest pain, 11 patients (61%) had dyspnea, and 3 patients (17%) had both. Time from symptom onset to MRI did not show a significant correlation to the number of AHA segments with alterations in mapping sequences or LGE. In the COVID-19 group, the maximum hsTnT showed a correlation to the number of AHA segments with abnormal T1 time (R2=0.602; P=0.011). The other imaging parameters of myocarditis did not show a correlation to maximum hsTnT or hsTnT at the time of CMRI. Six patients (33%) in the COVID-19 group had known CAD, all with no relevant stenosis at the time of CMRI and no alterations in CMRI that are typical for ischemic heart disease. Two patients had undergone chemotherapy with Gemcitabine.
TABLE 1 -
|Maximum HsTnT (ng/mL)
|HsTnT at MRI (ng/mL)
The mean cardiac parameters as depicted in Table 2 did not show any difference between both groups and were in normal ranges. Two patients in the COVID-19 group had markedly impaired LVEF of <30%; all others had an LVEF over 53%. In the control group, 3 patients had impaired LVEF of 20%, 34%, and 40%, respectively. Seven patients (39%) in the COVID-19 group had focal wall motion abnormalities compared with 6 patients (33%) in the control group. There was no significant difference in the number of patients with systolic dysfunction (5 vs. 3; P=0.691).
TABLE 2 -
|Ejection fraction (%)
|Indexed end-diastolic volume (mL/m2)
|Indexed end-systolic volume (mL/m2)
|Indexed stroke volume (mL/m2)
|Indexed myocardial mass (g/m2)
Cardiac parameters as determined with CMRI for both groups.
There was no significant difference between patients with any abnormalities in CMRI (15 [83%] vs. 18 [100%]; P=0.229). T2Q was significantly lower in patients with COVID-19 (1.6±0.6 vs. 2.4±0.6; P=0.008). T1, T2, and ECV showed no significant difference between the 2 groups (50.6±0.3 ms vs. 49.9±1.6 ms; P=0.095, 1045±58 ms vs. 1075±55 ms; P=0.129 and 30.4±8.1% vs. 29.4±5.8%; P=0.684, respectively).
Patients with global abnormal T1 in the COVID-19 group had significantly higher T1 than patients with only regional involvement (1,027±27 ms vs. 1,105±48 ms; P=0.004). No significant difference was observed for ECV (35.2±4.7% vs. 30.0±6.7%; P=0.152) or T2 (52.1±3.8 ms vs. 48.5±2.9%; P=0.77) when comparing these groups. Significantly fewer patients in the COVID-19 group had LGE and pericardial effusion (12 vs. 17; P=0.019 and 4 vs. 15; P=0.001, respectively).
In terms of LLC, there was no significant difference in the presence of NIMI (15 [83%] vs. 18 [100%]; P=0.229), but there was a significant difference in the number of patients with ME (7 [39%] vs. 15 [83%]; P=0.015) and who fulfilled LLC criteria (6 [33%] vs. 17 [94%]; P<0.001).
When comparing patients with and without LLC in the COVID-19 group, patients who fulfilled LLC had significantly higher T2 (52.0±3.1 ms vs. 47.8±2.4 ms; P=0.010) and higher T2Q (2.3±0.3 vs. 1.2±0.3; P<0.001). The mean T1 and ECV were not significantly elevated in LLC-positive patients (1,061±74 ms vs 1038±452 ms; P=0.460 and 35.9±4.3% vs. 28.1±8.3%; P=0.068, respectively). Detailed findings regarding LLC are depicted in Table 3.
TABLE 3 -
Lake Louise Criteria
| T2 mapping
| Fat-saturated T2 imaging
|Nonischemic myocardial injury
| T1 mapping
| Extracellular volume
| Gadolinium late enhancement
|Lake Louise criteria met
| Pericardial effusion
| Systolic LV dysfunction
Pathologic alterations and number of individuals fulfilling the Lake Louise Criteria for myocarditis.
In the COVID-19 group, EDVI showed a significant correlation to the number of AHA segments with alterations in T1 (R2=0.571; P=0.017) and ECV (R2=0.605; P=0.013) and mean T1, T2, and T2Q (R2=0.644; P=0.005, R2=0.513; P=0.035, and R2=0.629; P=0.038, respectively). These correlations are depicted in Figures 2 and 3. In the control group, only the mean ECV showed a weak positive correlation to EDVI (R2=0.490; P=0.046). There was no significant correlation of any CMRI parameter to EF in both groups.
In this single-center study, we compared CMRI morphologic parameters in patients with confirmed COVID-19 and elevated hsTnT to patients with suspected myocarditis.
We observed cardiac alterations in 83% of patients with COVID-19, whereas 33% of patients fulfilled the LLC for myocarditis. In the control group, all patients showed abnormalities and 94% fulfilled LLC.
Parameters of ME—T2 and T2Q—were less common in the COVID-19 group than in the control, while there was no significant difference in T1 or ECV. A possible explanation for this could be found in the case report of Tavazzi et al,8 who detected SARS-CoV-2 in an EMB from a patient with cardiogenic shock through electron microscopy, whereas light microscopy showed only low-grade inflammation. Similar results were also reported by Weckbach et al,16 who found typical alterations for lymphocytic myocarditis in only 1 patient and highly elevated macrophage numbers in another 4 patients with active COVID-19. The number of patients with pericardial effusion was also lower in the COVID-19 group as a further sign of less inflammation.
Despite the low ME, both the COVID-19 and control groups had similarly abnormal mean ECV of >28%, pointing to diffuse NIMI. In contrast, the control group showed more LGE. The mean T1 was lower in the COVID-19 group without reaching statistical significance, which is plausible, as it is a combined parameter indicating both NIMI and ME.17
Two patients in the COVID-19 group and 3 patients in the control group had impaired LV function. In 1 patient, mapping sequences were not performed, but the typical subepicardial inferolateral LGE pattern and diffuse ME in T2wfs were observed. The second patient had diffuse elevation of ECV, but no evidence of ME. The fact that most patients had normal LVEF could be attributed to observations that myocardial alterations in MRI might precede alterations in cardiac function.18
In patients with COVID-19, individual LLC parameters, both in their extent (T1 and ECV) and in their magnitude (mean T1, T2, and T2Q), showed a significant correlation to LVEDVI, a parameter of LV-dilation. In the control group, only ECV showed a weak correlation, whereas LVEDVI showed no significant difference between both groups. Both findings could lead to future cardiovascular complications. On the one hand, dilated cardiomyopathy is well associated with congestive heart failure and arrhythmias.19 On the other, in a recent study, NIMI was shown to have a very strong association with sudden cardiac death in nonischemic cardiomyopathy.20
Huang et al21 reported 58% cardiac involvement in a cohort of patients who had recovered from COVID-19 with cardiac symptoms. In contrast to our study population, their results showed ME as a predominant imaging feature, whereas only 57% of their examined patients showed NIMI. The most plausible reason for this might be that they only recruited patients with cardiac symptoms, whereas we examined all available patients with a modest increase in hsTnT, detecting myocardial alterations in patients without specific cardiac symptoms. Another recent study examining patients who had recovered from COVID-19 reported a similar prevalence of abnormalities in CMRI (78% of patients), but showed a positive correlation of both T1 and T2 to hsTnT compared with only T1 in this study.10
As COVID-19 can also lead to a hypercoagulatory state, it has been described as a possible cause for acute coronary syndrome (ACS).7 As none of our patients showed an ischemic pattern in CMRI and had no significant CAD, we decided to select patients with myocarditis as a control group for this study. This choice can be further underlined by a recent systematic review that reported ACS as a rare cardiac complication in COVID-19, adding up to <5% of cardiac complications detected by CMRI in 199 patients.22 Unfortunately, there are no details of the presence and degree of CAD in this study to establish whether these patients had an ACS independent of COVID-19 or coronary thrombosis due to a hypercoagulatory state.23
The main limitation of our study is the small sample size. This is primarily due to the small number of patients with COVID-19 who were treated at our hospital (116 in total, 73 with elevated hsTnT during the recruitment period). We examined as many patients as during the current situation as allowed. Some eligible patients were not able to perform the required breath-hold needed for CMRI; thus, the examination was not performed for these patients. We have no previous CMRI or follow-up examinations; thus, causality cannot be proven. The low sample size might also have had an influence on the low number of patients with ME in the COVID-19 group.
2 patients had received previous treatment with gemcitabine, in which cardiotoxicity is a rare complication. In a thorough literature research, we could find only one case series that, unfortunately, offered no reference to alterations in CMRI.24 One of these patients had myocarditis-typical inferolateral subepicardial LGE and the other had no abnormalities. In other known chemotherapy treatments, cardiac alterations in CMRI manifest as diffuse fibrosis25; thus, it seems unlikely that the alterations in the one patient are due to previous gemcitabine treatment.
Age and sex differed significantly between the groups, which could have an influence on functional parameters and T1/T2 values.
There were a few incomplete examinations missing either mapping sequences or LGE. Although some information was missing, the present sequences were sufficient to cover both NIMI and ME criteria of LLC for myocarditis.
In conclusion, myocardial involvement in COVID-19 shows a different phenotype in CMRI compared with all-cause myocarditis. More extensive cardiac involvement seems to lead to LV dilation and there seems to be less ME in patients with COVID-19. We believe that patients with COVID-19 and NIMI should be followed up closely for future complications even if they do not fulfill the traditional LLC for myocarditis. These initial findings in a small cohort should be further explored in larger studies.
1. Dong E, Du H, Gardner L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect Dis. 2020;20:533–534.
2. Shi H, Han X, Jiang N, et al. Radiological findings from 81 patients with COVID-19 pneumonia in Wuhan, China: a descriptive study. Lancet Infect Dis. 2020;20:425–434.
3. Pollack A, Kontorovich AR, Fuster V, et al. Viral myocarditis
—diagnosis, treatment options, and current controversies. Nat Rev Cardiol. 2015;12:670–680.
4. Shi S, Qin M, Shen B, et al. Association of cardiac injury with mortality in hospitalized patients with COVID-19 in Wuhan, China. JAMA Cardiol. 2020;5:802–810.
5. Akhmerov A, Marbán E. COVID-19 and the heart. Circ Res. 2020;126:1443–1455.
6. Chen C, Zhou Y, Wang DW. SARS-CoV-2: a potential novel etiology of fulminant myocarditis
. Herz. 2020;45:230–232.
7. Dominguez-Erquicia P, Dobarro D, Bastos-Fernandez G, et al. Multivessel coronary thrombosis in a patient with COVID-19 pneumonia. Eur Heart J. 2020;41:2132.
8. Tavazzi G, Pellegrini C, Maurelli M, et al. Myocardial localization of coronavirus in COVID-19 cardiogenic shock. Eur J Heart Fail. 2020;22:911–915.
9. Escher F, Pietsch H, Aleshcheva G, et al. Detection of viral SARS-CoV-2 genomes and histopathological changes in endomyocardial biopsies. ESC Heart Fail. 2020;7:2440–2447.
10. Puntmann VO, Carerj LM, Wieters I, et al. Outcomes of cardiovascular magnetic resonance imaging in patients recently recovered from coronavirus disease 2019 (COVID-19). JAMA Cardiol. 2020;5:1265–1273.
11. Ferreira VM, Schulz-Menger J, Holmvang G, et al. Cardiovascular magnetic resonance in nonischemic myocardial inflammation: expert recommendations. J Am Coll Cardiol. 2018;72:3158–3176.
12. Lessick J, Ghersin E, Abadi S, et al. Accuracy of the long-axis area-length method for the measurement of left ventricular volumes and ejection fraction using multidetector computed tomography. Can J Cardiol. 2008;24:685–689.
13. Messroghli DR, Moon JC, Ferreira VM, et al. Clinical recommendations for cardiovascular magnetic resonance mapping of T1, T2, T2* and extracellular volume: a consensus statement by the Society for Cardiovascular Magnetic Resonance (SCMR) endorsed by the European Association for Cardiovascular Imaging (EACVI). J Cardiovasc Magn Reson. 2017;19:75.
14. Haaf P, Garg P, Messroghli DR, et al. Cardiac T1 mapping and extracellular volume (ECV) in clinical practice: a comprehensive review. J Cardiovasc Magn Reson. 2016;18:89.
15. White JA, Hansen R, Abdelhaleem A, et al. Natural history of myocardial injury and chamber remodeling in acute myocarditis
. Circ Cardiovasc Imaging. 2019;12:e008614.
16. Weckbach LT, Curta A, Bieber S, et al. Myocardial inflammation and dysfunction in COVID-19-associated myocardial injury. Circ Cardiovasc Imaging. 2021;12:e012220.
17. Ugander M, Bagi PS, Oki AJ, et al. Myocardial edema as detected by pre-contrast T1 and T2 CMR delineates area at risk associated with acute myocardial infarction. JACC Cardiovasc Imaging. 2012;5:596–603.
18. Sangaralingham SJ, Huntley BK, Martin FL, et al. The aging heart, myocardial fibrosis and its relationship to circulating C-type natriuretic peptide. Hypertension. 2011;57:201–207.
19. Schultheiss HP, Fairweather D, Caforio ALP, et al. Dilated cardiomyopathy. Nat Rev Dis Primers. 2019;5:32.
20. Klem I, Klein M, Khan M, et al. The relationship of LVEF and myocardial scar to long-term mortality risk and mode of death in patients with non-ischemic cardiomyopathy. Circulation. 2021;143:1343–1358.
21. Huang L, Zhao P, Tang D, et al. Cardiac involvement in patients recovered from COVID-2019 identified using magnetic resonance imaging. JACC Cardiovasc Imaging. 2020;13:2330–2339.
22. Ojha V, Verma M, Pandey NN, et al. Cardiac magnetic resonance imaging in coronavirus disease 2019 (COVID-19): a systematic review of cardiac magnetic resonance imaging findings in 199 patients. J Thorac Imaging. 2020;36:73–83.
23. Knight DS, Kotecha T, Razvi Y, et al. COVID-19: myocardial injury in survivors. Circulation. 2020;142:1120–1122.
24. Khan MF, Gottesman S, Boyella R, et al. Gemcitabine-induced cardiomyopathy: a case report and review of the literature. J Med Case Rep. 2014;8:220.
25. Neilan TG, Coelho-Filho OR, Shah RV, et al. Myocardial extracellular volume by cardiac magnetic resonance in patients treated with anthracycline-based chemotherapy. Am J Cardiol. 2013;111:717–722.