Introduction

The coronavirus disease of 2019 (COVID-19) was first identified in China at the end of the 2019, resulting global pandemic and caused more than 102M cases and 2.2M related death worldwide [1, 2]. COVID-19 has variable clinical manifestations ranging from asymptomatic infection to severe acute respiratory failure, multi-organ dysfunction, and death [3]. COVID-19-related myocardial injury, including myocarditis, myocardial infarction, stress cardiomyopathy, heart failure, arrhythmias, and cardiogenic shock is an increasingly recognized complication seen in 8 to 12% of all patients [4]. COVID-19-related myocardial injury, including myocarditis, myocardial infarction, stress cardiomyopathy, heart failure, arrhythmias, and cardiogenic shock, is an increasingly recognized complication seen in 8 to 12% of all patients. Myocardial injury, including myocarditis, myocardial infarction, stress cardiomyopathy, heart failure, arrhythmias, sepsis-related cardiac injury, and cardiogenic shock, have been reported [5,6,7]. Several mechanisms have been hypothesized, including a pro-inflammatory state caused by cytokine storm, direct viral invasion of cardiomyocytes, a thromboembolic event caused by a hypercoagulable state, atherosclerotic coronary plaque instability, or rupture has been suggested [8,9,10,11].

The imaging of myocardial involvement is challenging due to the heterogeneity of the disease and multi-organ involvement. Endomyocardial biopsy is accepted as the gold-standard technique. However, its application in routine clinical practice is limited because of the periprocedural complications and technical difficulties [12]. Non-invasive imaging modalities provide a safe alternative for diagnosing and managing the treatment. Transthoracic echocardiography is the first-choice imaging technique due to its low cost, easy accessibility, and shorter imaging time [13]. Observer dependant assessment, limited ability in tissue characterization, and acoustic window are the main disadvantages. Cardiac magnetic resonance imaging (MRI) is a widely accepted gold standard non-invasive imaging tool because of its unique capability for non-invasive tissue characterization and mapping, which is recommended by multiple statements and reviews for the diagnosis of myocardial disease [14,15,16,17]. Myocardial mapping techniques, including native T1/post-contrast T1 mapping (extracellular volume mapping) for fibrosis, T2 mapping for edema, and late gadolinium enhancement for necrosis or fibrosis are decision-making options for the diagnosis of myocardial inflammation and monitoring response the therapy [18, 19]. Unfortunately, the requirement for gadolinium-based contrast agents which is contraindicated in patients with renal failure and hypersensitivity is the main limitation of the technique [20]. Relatively long scanning time and requiring the precision of appropriate inversion time are other limitations of contrast-enhanced myocardial mapping and tissue characterization techniques [21]. Thus, an alternative imaging tool to assess myocardial disease without needing gadolinium-based contrast agents would be preferable in patients with multi-organ dysfunction and renal insufficiency.

T2* mapping is the current method of choice for the non-invasive assessment of cardiac iron deposition and the evaluation of the response to the treatment with chelation therapy and recommended imaging tool in practically all clinical guidelines relating to iron overload treatment [22]. For all that, recent studies have focused on T2* mapping for post-revascularization hemorrhage [23, 24], hypertrophic cardiomyopathy [25, 26], or myocardial fibrosis [21]. On the other hand, T2* mapping for non-COVID-19 myocarditis has become an area of great interest [27, 28]. Although several studies have been published about the myocardial T1 mapping in COVID-19 related myocardial injury[29] myocardial T2* mapping in COVID-19 has not been evaluated yet. Therefore, we aimed to assess myocardial T2* values in patients with COVID-19 infection without significant cardiac symptoms. To our knowledge, this is the first study that assessed myocardial T2* mapping in COVID-19.

Materials and methods

This prospective study was approved by the institutional research and ethics committee, and informed consent for cardiac MRI examination was obtained from all patients.

Patient selection

A total of 20 healthy subjects and 25 patients with COVID-19 pneumonia between January 2021 and March 2022 were included in the study. All the individuals in the patient group were within two-to-five weeks of hospitalization and diagnosed with COVID-19 by polymerase chain reaction on a swab test at the administration. All the patients had a moderate pulmonary infection. Exclusion criteria were a history of coronary artery disease, hypertension, diabetes, or myocarditis, claustrophobia, contradictions to the gadolinium contrast, MRI non-compatible implants, pacemakers, and cardiac MRI image quality that was not sufficient for analysis. Healthy control subjects without known cardiovascular disease, chest pain, palpitations, exercise intolerance, or systemic inflammation were included in the study. Patients younger than 18 years old were also excluded from the study.The individuals for the control group were retrospectively selected by review of the cardiac MRI database who previously underwent the cardiac MRI without any prior known cardiac or inflammatory disease with normal biventricular volume, function, and ECG. We confirmed that the individuals in the control group were negative for COVID-19 and cardiac complaints after vaccination during the examination. We recorded demographic and clinical data, including blood tests and cardiac enzymes, from the hospital health information system.

Cardiovascular ımaging and technique

All CMR examinations were performed using a 1.5-Tesla scanner (Explorer; GE healthcare systems, Chicago, IL, USA) with a 32-channel phased-array abdominal coil with electrocardiographic gating. The duration of the examination changes between 30 and 35 s. The weight and height of the patients were recorded before the examination and body surface area was calculated by the Mosteller Formula automatically. No gadolinium-based contrast agents or sedation were used.

All the examinations were performed following our institutional standardized cardiac MRI protocol with the imaging parameters below. After three plane localizers through the thorax were revealed, balanced steady-state free precession (bSSFP) cine images for two-chambers, four-chambers, and short axis plane covering the entire ventricles. The parameters of the bSSFP sequence were revealed with the following typical parameters: echo time;1.4 ms, repetition time; 37.7 ms, bandwidth; 780 Hz/pixel, pixel size; 2.2 × 2.2 mm, a field of view; 350 × 350 mm, matrix;192 × 146, flip angle; 55°, slice thickness;8 mm, and slice gap; 2 mm. The images were acquired during one or two breath-hold of 7 to 13 s duration, depending on the heart rate during the end-expiratory breath hold.

Native-T1 map and T2* sequences were performed in 3 short axis slices at the basal, mid, and apical short axis at the same imaging plane using a breath-held. Native-T1 mapping was acquired using an electrocardiograph-triggered single-shot modified Look-Locker inversion recovery (MOLLI) sequence during diastole using a 5(3)3 sequence. The acquisition parameters were: echo time; 1.5 ms, repetition time; 3.8 ms, bandwidth; 1250 Hz/pixel, pixel size; 2.2 × 2.2 mm, field of view; 320 × 360 mm, matrix; 192 × 144, flip angle; 35°, slice thickness; 8 mm, and acceleration factor; 2. The T2* mapping was acquired using a multi-TE gradient echo with different TE times; 3, 7, 10, 14, 17, 20, 23, 26, and 29 ms, TR; 13.2 ms, flip-angle; 20°, bandwidth; ± 83.33 kHz, matrix; 128 × 256, FOV; 350 × 350 mm, NEX; 1, slice thickness; 8 mm.

All the cardiac MRI analyses were performed using Circle cvi42 software (Circle Cardiovascular Imaging, Calgary, Canada) by ten years experienced radiologist (trained in cardiac MRI with an experience with more than 800 cardiac MRI examinations). The endocardial layer of both ventricles was contoured manually on short-axis cine images by including the papillary muscles and the trabeculations through all slices on end-diastolic and end-systolic phases to calculate ventricular volume and function. Ejection fraction, indexed end-diastolic and end-systolic volumes (EDV and ESV, respectively) were calculated automatically. Native-T1 and T2* mappings were traced in the respective CVI42 modules (Fig. 1). Endocardial and epicardial borders of the left ventricle were traced manually at the same 3 imaging planes in base, middle, and apical segments of left ventricle. Pericardial fat tissue and ventricular blood pool were not included in ROI measurements to avoid partial volume averaging artifacts at the same imaging plane.

Fig. 1
figure 1

Multiparametric cardiac magnetic resonance imaging in a 47-year-old male patient with COVID-19. The same slices of short axis view on basal, mid-ventricular, and apical segments (left-to-right) for b-SSPF, native-T1, and T2* mapping images. Native-T1 and T2* mapping values were calculated for the entire ventricle as 1028 ± 8 ms and 41 ± 6 ms, respectively

Full size image

Statistics

The analysis of the obtained data was assessed with SPSS version 20.0 software (IBM Corporation, Armonk, NY, USA). The data were expressed as the mean ± standard deviation (for normal distribution) or medians and interquartile ranges [IQR] (for non-normal distribution). The Shapiro–Wilk normality test was used to evaluate the normality of the obtained data. The Student’s T-test was used to compare the T2* mapping values and other variables between groups. Pearson conversion coefficient (r) for normal distribution data and Spearman conversion coefficient (rho) for abnormal data distribution was performed to evaluate of correlation between cardiac MRI mapping values and serum parameters. p < 0.05 was considered statistically significant for all the comparisons.

Results

Twenty-five patients with COVID-19 (14 females and 11 males, mean age of 41.76 ± 10.19 years, median age: 40 years, [IQR 35–49 years] and 20 healthy individuals (9 females and 11 males, mean age of 42.10 ± 9.16 years, median age: 44 years [IQR 35–49 years]) were included to the study (p = 0.908). There was no statistically significant difference between the patient and control groups in terms of gender (p = 0.65) or body surface area (patient body surface area 1.92 ± 0.23m2 vs. control body surface area 11.99 ± 0.22 m2, p = 0.299). The laboratory findings of patients with reference values are presented in Table 1. Concerning cardiac MRI assessment, there was no difference between patient and control groups in terms of left ventricular ejection fraction (55.72 [IQR 51.54–59.05% vs. 56.02 [IQR 52.22–59.53] %; p = 0.712), left ventricular end-diastolic volume index (74.54 [IQR 61.99–86.32] ml/m2 vs. 55.15 [IQR 69.67–87.16] ml/m2; p = 0.536), right ventricular ejection fraction (52.12 [IQR 46.48–56.77] % vs. 53.11 [IQR 48.16–58.17] %; p = 0.796), and right ventricular end-diastolic volume index (68.05 [IQR 60.71–80.57] ml/m2 vs. 73.96 [IQR 66.10–81.55] ml/m2; p = 0.293). T2* mapping values were lower in patients with COVID-19 compared to healthy controls (35.37 [IQR 31.67–41.20] ms vs. 43.98 [IQR 41.97–46.88] ms; p < 0.0001), while there was no significant difference in terms of native-T1 mapping values (1088.02 ms [IQR 1042.83–1140.83 ms] vs. 1073.07 ms [IQR 1044.37–1098.01 ms]; p = 0.701). Cardiac MRI parameters of both groups are presented in Table 2.

Table 1 Serum biochemical characteristics of the patients with COVID-19 and control groups
Full size table
Table 2 Cardiac MRI characteristics of the patients with COVID-19 and control groups
Full size table

There was a positive correlation with T2*mapping and native-T1 mapping values (r = 0.522, p = 0.007) and negative correlation with serum ferritin values (r = -0.653, p = 0.000) (Fig. 2) (Table 3). However, there was no correlation between cardiac T2* mapping and serum troponin-I, d-dimer values (r = 0.434, p = 0.056 and r = 0.092, p = 0.692, respectively) (Table 3). There was no correlation between cardiac native-T1 mapping and serum ferritin, troponin, d-dimer values (r =  − 0.151, p = 0.470, r = 0.043, p = 0.857, and r = -0.089, p = 0.700, respectively) (Fig. 3) (Table 3).

Fig. 2
figure 2

Correlation of mean T2* mapping value with serum ferritin (a, left) and native-T1 mapping (b, right) values

Full size image
Table 3 Correlation between serum parameters with myocardial native-T1 map and T2* map values
Full size table
Fig. 3
figure 3

Correlation between mean native-T1 mapping value with serum ferritin

Full size image

Discussion

In this study, the T2* mapping value was found to be significantly higher in patients with COVID-19 compared to healthy subjects. However, no significant difference was found between two groups in terms of native-T1 mapping value. There was a significantly negative correlation between T2* mapping and serum ferritin level in patients with COVID-19 infection and a positive correlation between T2*and native-T1 mapping values. However, no significant correlation was found between the native-T1 mapping value and serum ferritin level.

T2* mapping is able to characterize the relaxation of the transverse magnetization that is influenced by magnetic field inhomogeneities composed of two elements. One is spin–spin interaction effect that is caused by its own inhomogeneities, terms as a macroscopic effect. The second one is the local magnetic field susceptibility changes caused by the structure of the tissue as a mesoscopic effect [15, 30]. Thus, the T2* relaxation time is a more sensitive measure to evaluate the structural and pathological changes in tissue [24]. T2* mapping assessment of microvascular obstruction, acute myocardial infarction, and intramyocardial hemorrhage [23, 24, 31] has become of great interest in recent years. Hemoglobin aggregation within infarcted areas of the myocardium causes signal nulling [31]. Iron shortens T2* relaxation times and can be measured by T2* mapping[32].

The exact mechanisms of myocardial injury in COVID-19 have not been completely understood. Most likely mechanisms are suggested to be direct virus-mediated cytotoxicity, dysregulation of the renin–angiotensin–aldosterone system, endothelial cell damage, thrombo-inflammation, dysregulation of the immune response, and hypoxic injury [12]. It is suggested that the endothelium has a crucial role in the pathogenesis of severe COVID-19, inducing both pro-coagulant and inflammatory responses [33, 34].In addition to the direct cytopathic viral effect, severe COVID-19 infection would involve both large and microscopic blood vessels. It can be complicated by infection-induced microangiopathy or thromboembolic disease that causes capillary, venous and/or arterial thrombosis and end-organ damage [35]. It is also mentioned that microthrombi is to be associated with myocyte necrosis in autopsy studies [36]. Additional several ischemic mechanisms, including type-2 myocardial infarction and non-ischemic mechanisms, such as microangiopathic thrombosis, have also been proposed to be responsible [37]. Considering the possibility of hemorrhages, thrombi, or dysregulation of iron metabolism, we hypothesized that measurement of iron deposition through myocardial T2* mapping without using contrast agents injection could be used to assess myocardial changes in patients with COVID-19.

T2* mapping for non-COVID-19 myocarditis has been assessed in two MRI studies [27, 28]. Helluy et al. [27] assessed myocardial injury in murine models of acute and chronic viral myocarditis by T2* mapping [27]. Although quantitative assessment has not been performed and T2* mapping values not computed, direct localization of virus-induced myocardial lesions ex vivo and in vivo was demonstrated by T2* imaging without any MRI tracer or contrast agent [27]. It is also mentioned that T2* mapping has the capability to detect both small cardiac lesions of acute myocarditis and larger necrotic areas at later stages of chronic myocarditis, which was correlated histopathologically [27]. In another study by Baxan et al. [28]] assessed acute hemorrhagic myocarditis in ex vivo histopathologic correlation with in vivo cardiac MRI using T2* sequence in the animal model and suggested that low T2 *detecting areas of increased levels of iron in the tissue suggestive of inflammatory changes including vascular damage, red blood cell extravasation, and tissue iron deposition.

Hyperferritinemia in patients with COVID-19 infection and the association of high ferritin levels with the severity of the illness and mortality have been investigated early in the course of the pandemic [38]. Dysregulated iron metabolism is suggested to be responsible in severe COVID-19 disease [38, 39]. Although the exact mechanisms of ferroptosis and the association between iron are not clearly identified yet, it is well known the crucial role of iron metabolism in ferroptosis, which is a recently identified type of controlled cell death leading to excessive reactive oxygen species formation, lipid peroxidation triggered by intracellular iron accumulation, decreased iron storage capacity and glutathione depletion [39, 40].

In addition to reducing the need for myocardial biopsy, in vivo and ex vivo experiments have made significant advances in cardiac MRI, which is now considered the clinical gold standard for non-invasive detection of myocardial damage and structural changes. Researches about COVID-19 and its cardiac effects have recently focused on in vitro and ex vivo studies [39, 41]. An in vivo study by Sharma et al. [41] demonstrated that COVID-19 could infect and induce apoptosis in human induced pluripotent stem cell-derived cardiomyocytes. Although ferroptosis is not investigated in this study, it is an essential point that myocarditis is caused by a direct viral infection instead of immune cell infiltration. In another study by Baier et al. [39], iron metabolism and association with ferroptosis were investigated in the left ventricular myocardium of patients who had died due to severe COVID-19. They found that increased intracellular iron concentration is associated with the inhibition of detoxifying enzymes (GPX4) as well as activation of the ferroptosis pathway (HMOX-1), leading to increased myocardial oxidative stress with subsequent myocardial damage. Although a direct relationship between cardiac iron overload and cardiac injury in COVID-19 infection was shown, the study lacks cardiac imaging and clinical correlation to corroborate the postmortem findings. Dysregulation of iron metabolism and its role in neurological, renal, circulatory, and immune system regulation is also implicated [39]. This multisystem involvement would explain the heterogeneous clinical symptoms and multiple organ involvement in COVID-19.

Another study by Han Y. et al. has defined that sinoatrial pacemaker cells are infected directly by the COVID-19 virus and induced ferroptosis [42]. Another important finding of this study is that the treatment of deferoxamine which is used as a chelator of iron overload and known to be an inhibitor of ferroptosis, blocked the viral entry of COVID-19 and inhibited ferroptosis. The effect of iron chelators is seen in multiple steps, such as chelating iron, inhibiting redox properties exerted by free iron, and hindering the participation of iron in Fenton reactions which allows inhibition of hydroxyl radical production as well as the creation of other ROS that protect oxidative stress and ferroptosis [43]. It is also shown that iron chelators block ferroptotic cell death in vitro and in vivo [43]. Iron chelators have been used to treat cytokine storms and multi-organ injuries in a variety of environments, including ischemia–reperfusion injury, sepsis, and infections, and have been shown to help with several viral infections, including HIV-1, hepatitis B, and mengovirus infection, among others [44]. Prospective studies in human and animal models are needed to fully elucidate the efficacy and safety of iron-chelating agents in the therapeutic armamentarium of COVID-19, probably as an adjunctive treatment.

Our result (negative correlation between cardiac T2* mapping and serum ferritin) indicates that increased ferritin level would decrease myocardial T2* mapping values in COVID-19. We suggest that a negative correlation between myocardial T2* mapping value and ferritin level would be suggestive of microhemorrhages in the macroscale aspect or ferroptosis-mediated myocardial cell destruction in the microscale aspect. Although it is known that iron shortens both T1 and T2* relaxation times [32], no significant difference was found between the control and patient groups in terms of native-T1 mapping in our study. Recent studies have demonstrated that T1 mapping with prolonged T1 relaxation time is a useful diagnostic tool in the assessment of fibrosis and interstitial remodeling of the myocardium [45]. However, the current evidence conflicts with recent studies reporting decreased T2* values in hypertrophic cardiomyopathy [26, 46] or in post-myocardial infarction [21], especially those who had myocardial fibrosis or fibrotic segments. The result of our study with decreased T2* value without increased native-T1 value is speculative. Regarding this finding, we hypothesize that potential discrepancies in oxygen delivery to the myocardial tissue or a complex interaction between fibrosis and myocardial inflammation might be involved. It is suggested that reduced T2* mapping values in patients with hypertrophic cardiomyopathy are potentially triggered by relative ischemia [21]. Another hypothesis is that decreased levels of oxyhemoglobin and oxymyoglobin in case of reduced perfusion may be associated with reduced T2* values because of the paramagnetic effect of deoxygenated proteins [46]. On the other hand, reduced T1 mapping value is reported at follow-up examinations 68 days after the baseline scan [47]. Depending on the severity of the microhemorrhage or hypoperfusion and the regional/segmentary or diffuse involvement, there might be either an improper immune response with persistent inflammation [16, 47] or a more subtle and diffuse process that drives the changes after COVID-19. Concerning the correlation between cardiac MRI measurements and serum parameters, there was no correlation between cardiac MRI mapping and serum troponin-I, d-dimer values. Previous studies have shown that high troponin levels are associated with increased mortality in patients with COVID-19. However, abnormal troponin levels are not necessarily a sign of an acute course of myocarditis [4].

The current study has a number of limitations. First, the study cohort is limited. A multicenter study with a larger patient group would strengthen the results. Second, the absence of late gadolinium-enhanced imaging or CMR perfusion imaging and verifying the relation between T2* and fibrosis. However, studies from different tissues have already shown the value of native-T1 mapping for detecting fibrosis [24]. Cardiac MRI also has its limitations, such as higher cost, longer scanning time, and the need for trained specialists.

In conclusion, the negative correlation between serum ferritin level and T2* mapping values in COVID-19 patients may provide a non-contrast-enhanced alternative to assess tissue structural changes in patients with COVID-19. Adding T2 * cardiac MRI mapping in patients with myocardial pathologies would improve reveal underlying mechanisms without contrast agent injection. In vivo and ex vivo animal or human studies designed with larger patient cohorts and correlation with histopathological findings should be planned in the future.