Author + information
- Received May 11, 2017
- Revision received July 19, 2017
- Accepted August 8, 2017
- Published online October 9, 2017.
- Giovanni Donato Aquaro, MDa,∗ (, )
- Matteo Perfetti, MDa,
- Giovanni Camastra, MDb,
- Lorenzo Monti, MDc,
- Santo Dellegrottaglie, MDd,e,
- Claudio Moro, MDf,
- Alessia Pepe, MDa,
- Giancarlo Todiere, MDa,
- Chiara Lanzillo, MDg,
- Alessandra Scatteia, MDh,
- Mauro Di Roma, MDi,
- Gianluca Pontone, MDj,
- Martina Perazzolo Marra, MD, PhDk,
- Andrea Barison, MDa,
- Gianluca Di Bella, MD, PhDl,
- on behalf of the Cardiac Magnetic Resonance Working Group of the Italian Society of Cardiology
- aGabriele Monasterio Foundation, Tuscan Region, Pisa, Italy
- bCardiac Department, Vannini Hospital Rome, Rome, Italy
- cRadiology Department, Humanitas Research Hospital, Hospital Care and Research Institution (IRCCS), Rozzano, Milan, Italy
- dDivision of Cardiology, Villa dei Fiori, Acerra, Naples, Italy
- eMount Sinai School of Medicine, New York, New York
- fDepartment of Cardiology and Coronary Intensive CareUnit, ASST Monza, Desio Hospital, Desio Monza e Brianza, Italy
- gCardiology Department, Casilino Polyclinic, Rome, Italy
- hDepartment of Advanced Biomedical Sciences, Federico II University, Naples, Italy
- iRadiological Department, European Hospital, Rome, Italy
- jCardiac Department, Monzino Cardiology Center, Milano, Italy
- kDivision of Cardiology, Department of Cardiac, Thoracic, and Vascular Sciences, University of Padua, Padua, Italy
- lClinical and Experimental Department of Medicine, University of Messina, Messina, Italy
- ↵∗Address for correspondence:
Dr. Giovanni Donato Aquaro, Fondazione Toscana G. Monasterio, Via Giuseppe Moruzzi, 1, 56124 Pisa, Italy.
Background The prognostic role of cardiac magnetic resonance (CMR) and late gadolinium enhancement (LGE) has not been clarified in acute myocarditis (AM) with preserved left ventricular (LV) ejection fraction (EF).
Objectives This study sought to evaluate the role of CMR and LGE in the prognosis of AM with preserved LVEF.
Methods This study analyzed data from ITAMY (ITalian multicenter study on Acute MYocarditis) and evaluated CMR results from 386 patients (299 male; mean age 35 ± 15 years) with AM and preserved LVEF. Clinical follow-up was performed for a median of 1,572 days. A clinical combined endpoint of cardiac death, appropriate implantable cardioverter-defibrillator firing, resuscitated cardiac arrest, and hospitalization for heart failure was used.
Results Among the 374 patients with suitable images, LGE involved the subepicardial layer inferior and lateral wall in 154 patients (41%; IL group), the midwall layer of the anteroseptal wall in 135 patients (36%; AS [anteroseptal] group), and other segments in 59 patients (16%; other-LGE group), and it was absent in 26 patients (no-LGE group). The AS group had a greater extent of LGE and a higher LV end-diastolic volume index than other groups, but levels of inflammatory markers were lower than in the other groups. Kaplan-Meier curve analysis indicated that the AS group had a worse prognosis than the other groups (p < 0.0001). Finally, in multivariable analysis, AS LGE was the best independent CMR predictor of the combined endpoint (odds ratio: 2.73; 95% confidence interval: 1.2 to 5.9; p = 0.01).
Conclusions In patients with AM and preserved LVEF, LGE in the midwall layer of the AS myocardial segment is associated with a worse prognosis than other patterns of presentation.
In the last 10 years, cardiac magnetic resonance (CMR) has significantly improved diagnosis in patients with suspected acute myocarditis (AM). The clinical presentation of AM is heterogeneous, and it may begin as recent-onset heart failure, with arrhythmic events, or with infarct-like symptoms (1–3). Systolic function is usually preserved in patients with infarct-like presentation, who are considered to have a good prognosis. In the subsets of hemodynamically stable AM with preserved systolic function, endomyocardial biopsy may not be indicated (4), and CMR is considered the best noninvasive imaging modality for a definite diagnosis of AM.
CMR can detect signs of myocardial damage such as myocardial edema, hyperemia, and late gadolinium enhancement (LGE) in patients with preserved left ventricular (LV) ejection fraction (EF) (5). Preserved systolic function is a good predictor of survival in all cardiac disease, and previous studies demonstrated that impairment of EF in AM is the strongest predictor of worse prognosis (6,7). In many heart diseases, the presence of myocardial LGE is associated with an increased risk for adverse cardiovascular events in patients with preserved systolic function (8,9), but in AM, most patients have positive LGE (10).
In nonselected patients with AM, Mahrholdt et al. (11) demonstrated 2 main patterns of LGE presentation: 1 involving the subepicardial layer of the lateral wall and 1 involving the midwall layer of the anteroseptal wall. The latter was associated with ventricular remodeling and EF impairment at follow-up CMR. The main indication for CMR in AM is in hemodynamically stable patients with preserved EF. The aim of this multicenter study was to evaluate the prognostic role of LGE in a large sample of patients with hemodynamically stable AM and preserved EF.
ITAMY (ITalian multicenter study on Acute MYocarditis) is a multicenter investigation of the prognostic value of CMR in AM of the Cardiac Magnetic Resonance Working Group of the Italian Society of Cardiology. The ITAMY registry includes 415 consecutive inpatients with a definite CMR diagnosis of AM who were enrolled from January 2006 to January 2013 in 10 Italian hospitals. CMR was performed in patients with suspected AM and different clinical presentations: new onset of chest pain, dyspnea, or ventricular arrhythmic events (ventricular tachycardia, resuscitated cardiac arrest, or new-onset third-degree atrioventricular block).
In the current study, we excluded patients with heart failure or arrhythmic events at presentation, decreased systolic function (EF <50%), and hemodynamic instability (Figure 1). For the diagnosis of AM, we applied the diagnostic algorithm reported in Figure 2, which was modified from the algorithm of the European Society of Cardiology guidelines (3). Briefly, clinically suspected AM was diagnosed when symptomatic patients with chest pain (pericarditis or pseudoischemic pain) fulfilled 1 or more diagnostic criteria (new electrocardiographic [ECG] modification, elevated troponin, wall motion abnormalities with preserved LVEF at echocardiography) or in asymptomatic patients with 2 or more diagnostic criteria. A definite diagnosis of AM was then made in the event of 2 or more CMR Lake Louise criteria (myocardial edema, hyperemia, and LGE) (5).
Endomyocardial biopsy was performed when CMR results were inconclusive (≤1 CMR criterion). To exclude obstructive coronary artery disease, coronary artery angiography was performed on all patients with the exception of those younger than 30 years of age with a low risk of coronary artery disease. The final population included 386 patients (299 male; mean age 35 ± 15 years) with a CMR diagnosis of stable AM and preserved systolic function (LVEF >50%). At hospital admission, all patients underwent clinical evaluation and laboratory testing, including leukocytes, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and troponin T. Informed consent was obtained from all patients at the time of CMR examination.
CMR acquisition protocol
CMR imaging was performed with 1.5-T systems (CVi, HD release, GE Healthcare, Milwaukee, Wisconsin; Magnetom Avanto, Siemens Medical Systems, Erlangen, Germany; Gyroscan NT, Philips Healthcare, Amsterdam, the Netherlands) using dedicated cardiac software, a phased-array surface receiver coil, and vectocardiogram triggering. According to the protocols recommended by the Society for Cardiovascular Magnetic Resonance, we acquired cine steady-state free precession (cine-SSFP) images, T2-weighted imaging, and LGE at 10 min after gadolinium injection in the short-axis (9 to 13 images covering the entire LV), 2-chamber, and 4-chamber planes. Short-axis cine-SSFP images were acquired immediately after gadolinium injection for hyperemia assessment.
All CMR studies were analyzed off-line using a workstation with dedicated cardiac software with consensus among 3 experienced observers who were blinded to the clinical presentation results. To evaluate LV global and regional function and calculate LV mass, the endocardial and epicardial borders were manually drawn in the end-diastolic and end-systolic short-axis cine-SSFP images. Papillary muscles and trabeculae were not included in the myocardium. LV end-diastolic volume (EDV), LV end-systolic volume, LVEF, and LV mass were determined.
On T2-weighted images, edema was considered present when the ratio of signal intensity between the myocardium and the mean signal intensity of the skeletal muscle was ≥2 (5,12). LGE was qualitatively evaluated and manifested as a nonischemic pattern of distribution (i.e., subepicardial or midventricular enhancement) (13,14). Myocardial hyperemia was evaluated, as previously reported (13), using the post-contrast SSFP cine images. The occurrence of edema, hyperemia, or LGE was evaluated in each of the 17 LV segments (13,14). Furthermore, each LV wall (septal, anterior, inferior, and lateral) was considered to be involved if at least 1 segment with LGE was observed (15).
After the CMR examination, follow-up was performed for all patients by the investigators of the 10 Italian hospitals in the registry. A clinical questionnaire was compiled by a clinical physician during periodic ambulatory visitations in each hospital, by contacting the patients’ relatives by telephone, by a general practitioner, or by consulting the office of vital statistics at a patient’s place of residence. The clinical questionnaire included the definition of the following major events: cardiac death, resuscitated cardiac arrest, ventricular assist device, transplantation, and appropriate implantable cardioverter-defibrillator (ICD) shock, as well as minor events (heart failure hospitalization). A complete analysis of the ICD was performed by the referring physician to confirm the appropriateness of the shock.
Values are presented as the mean ± SD or as the median (interquartile range [IQR]) for variables with normal and non-normal distributions, respectively. Values with non-normal distribution according to the Kolmogorov-Smirnov test were logarithmically transformed for parametric analysis. Qualitative data are expressed as percentages.
Categorical variables were compared by the chi-square test or the Fisher exact test when appropriate. Continuous variables were compared by the Student independent t test and analysis of variance or by the Wilcoxon nonparametric test when appropriate. The Kaplan-Meier time-to-event method was used to calculate and compare longitudinal curves among groups. Logistic regression analysis was used to explore the impact of each significant variable in univariate analysis to predict the occurrence of cardiac events evaluated as a combined endpoint (cardiac death, appropriate ICD firing, resuscitated cardiac arrest, and hospitalization for heart failure). A p value lower than 0.05 was considered statistically significant.
Baseline characteristics of the entire population and groups are summarized in Table 1. The population was composed of young adults with a low prevalence of risk factors for coronary heart disease. The most commonly reported symptom was chest pain (95%), and fever was present in 58% of patients. All patients were in New York Heart Association functional class I because of the exclusion of patients with heart failure presentation. ECG abnormalities were found in 371 (96%) of the subjects (ST-segment elevation in 73%, ST-segment depression in 15%, and negative T-wave in 8%). Increased troponin T was found in all patients, and abnormal values of ESR or increased CRP were found in 385 patients (99%). Regional wall motion abnormalities were found in 82 patients by using echocardiography (21%).
Coronary artery angiography was performed in 365 patients (95%), who all showed no obstructive coronary artery disease (patients with obstructive disease were excluded from the study). Coronary artery angiography was not performed in 21 patients younger than 30 years of age (mean age of these patients was 22 ± 9 years), patients who had a low risk of coronary artery disease, and patients with signs of inflammation (fever preceding chest pain, increase of ESR and CRP, or signs of viral infection). The definite diagnosis of AM was made using endomyocardial biopsy in 18 patients (5%) with only 1 tissue abnormality in CMR. In the remaining patients, clinically suspected myocarditis was made confirmed by the evidence of 2 or more tissue abnormalities in CMR. The diagnostic criteria are shown in Table 2.
LV dilation was found in 33 patients (8%; 28 male) by using recently published reference values of normality (16). Right ventricular (RV) dilation was found in 42 patients (10%). Pericardial effusion was found in 48 patients (12%), and signs of pericardial inflammation (post-contrast enhancement) were found in 11 (3%). Regional wall motion abnormalities were found in 92 patients (22%). In T2–short tau inversion recovery (STIR) images, signs of myocardial edema were detected in 362 patients (94%). Hyperemia was found in 85 patients (23%).
In 12 cases, LGE images were of suboptimal quality as a result of artifacts. LGE images of 374 patients were visually analyzed, among which 4 main clusters of distribution were found: in 154 patients (41%), the subepicardial layer of the inferior and lateral segments was constantly involved with variable distribution in other segments, except the anteroseptal segments (IL cluster); in 135 patients (36%), the midwall layer of the basal anteroseptal wall was constantly involved with various other segments, except the inferior or inferolateral walls (AS cluster); and in 59 cases (16%), LGE images were positive but did not involve the inferolateral basal or the anteroseptal walls (other-LGE cluster). Finally, in 26 patients (7%), LGE images were negative (no-LGE cluster). On the basis of these clusters, the population was subdivided into 4 groups: the IL group, AS group, other-LGE group, and no-LGE group (Figure 1). Examples of LGE images from these groups are shown in Figures 3 and 4.
As evident in Table 3, patients in the AS group had lower ESR and CRP values than did other groups. In addition, the AS group had a higher LV EDV index (LVEDVi) and more myocardial segments with LGE than the IL group. LVEF and wall motion score index were not different among groups, but patients in the AS group had a lower RVEF than others. The no-LGE and other-LGE groups showed higher RVEF than other patients. The number of myocardial segments with edema was higher in the AS and IL groups than in the no-LGE and other-LGE patients. Myocardial edema and hyperemia had the same regional distribution of LGE.
During the median follow-up of 1,572 days (25th to 75th percentile: 1,122 to 2,923), ICDs were inserted in 6 patients (2 patients for episodes of nonvasovagal syncope associated with nonsustained ventricular tachycardia during 24-h ECG Holter monitoring and in 4 patients for evidence of sustained ventricular tachycardia during Holter monitoring). There were 30 patients who had >5% worsening of LVEF according to echocardiography during follow-up. The average decrease of LVEF in these 30 patients was 10 ± 5%. In 26 of these patients, coronary angiography was repeated during follow-up and confirmed the absence of obstructive coronary artery disease. The worsening of LV function was more prevalent in the AS group than in the IL group (20 vs. 10; p = 0.022).
As reported in Table 4, there were 8 major cardiac events (4 sudden cardiac deaths; 2 resuscitated cardiac arrests, 2 appropriate ICD firings) and 21 hospitalizations for heart failure. Major cardiac events occurred more frequently in the AS group than in the IL group (AS: 6; IL: 0; p < 0.01), whereas no differences were found with other groups (2 events in the other-LGE group and none in the no-LGE group). The AS group also had more frequent hospitalizations for heart failure than did the IL group (15 vs. 4; p = 0.004). Fifteen patients had recurrence of AM with clinical manifestation of chest pain, troponin increase, and evidence of 2 or more signal abnormalities in the CMR evaluation. The prevalence of recurrence was not significantly different among the groups (5 for AS, 6 for IL, and 4 for other-LGE).
Overall, the combined endpoint of major cardiac events and hospitalization for heart failure occurred in 21 patients (16%) in the AS group, 4 (3%) in the IL group, and 4 (5%) in other-LGE group, whereas no event occurred in the no-LGE group. In the Kaplan-Meier curve analysis, the AS group had worse event-free survival rates than did the other groups (Central Illustration). No prognostic differences were found among IL, other-LGE, and no-LGE groups. The characteristics of patients with and without cardiac events are summarized in Table 5. A logistic regression analysis was carried out using the combined endpoint as the dependent variable and the AS group and troponin peak as independent variables. Belonging to the AS group was the best predictor of the combined endpoint (AS group: odds ratio: 2.73; 95% confidence interval: 1.20 to 5.90; p = 0.01; and troponin peak: odds ratio: 1.14; 95% confidence interval: 1.03 to 1.63; p = 0.12).
The main results of the present study are as follows: 1) 3 main patterns of LGE were found in patients with AM and preserved LVEF: IL, AS, and other-LGE; 2) the AS group had higher a RVEDVi and LVEDVi, higher troponin release, and lower inflammatory markers; 3) the AS group was associated with a worse prognosis according to Kaplan-Meier analysis; and 4) being in the AS group was the best CMR predictor of the combined endpoint (Central Illustration).
We selected patients with AM, preserved LVEF, and New York Heart Association functional class I. Almost all these patients had an infarct-like presentation with chest pain and ECG abnormalities. Infarct-like myocarditis in patients with preserved EF is probably 1 of the most appropriate indications for CMR. As demonstrated by Francone et al. (17), CMR is very sensitive for identifying AM with infarct-like presentation compared with endomyocardial biopsy, but it is less accurate for the diagnosis of myocarditis when the initial presentation includes heart failure or arrhythmic events.
Although preserved systolic function is a good predictor of survival in many cases of heart disease, the presence of myocardial scar is generally associated with increased risk for adverse cardiovascular events, even in patients with preserved systolic function. In a large group of 857 patients with ischemic and nonischemic cardiomyopathies, Cheong et al. (8) showed that the presence of scar tissue expressed by LGE predicted a worse outcome than did the absence of LGE, even in patients with EF >50%. The data were confirmed in a cohort of 1,068 consecutive patients (9), in which LGE was further associated with a high occurrence of hospitalization for heart failure, regardless of etiology, stage of heart failure, or severity of EF impairment.
Grün et al. (7) studied 203 patients with a definite diagnosis of AM detected by endomyocardial biopsy and found that patients with LGE had larger ventricles, lower LVEF, and a worse prognosis. Particularly, these investigators observed cardiac death during follow-up in 28 of 29 of patients with positive LGE and only in 1 patient with negative LGE. However, all their patients with events had impaired EF.
In 2014, Schumm et al. (18) used CMR to evaluate 405 patients with suspected AM. These investigators found that patients with abnormalities at CMR (with a final diagnosis of AM or other cardiac diseases) had a worse prognosis than did patients with negative CMR findings. More recently, Sanguineti et al. (6) followed 203 patients with a CMR-based diagnosis of AM for an average of 18.9 months. These investigators observed that the presence and extent of myocardial edema and the extent of LGE were not predictive of the outcome. An impaired LVEF at the first examination was the only independent CMR predictor of an adverse clinical outcome (6).
In our multicenter study, we found that in AM with preserved LVEF, LGE had different patterns of presentation. The most frequent pattern found in 41% of cases involved the subepicardial layer of the inferior and lateral wall of the left ventricle (IL group). In 36% of patients, LGE was located in the midwall of the interventricular septum (AS group). The AS pattern of LGE was associated with a worse prognosis than in patients with the IL and no-LGE or other-LGE patterns. Patients with the AS pattern had no significant differences in LVEF from other groups, despite higher LVEDVi and more myocardial segments with LGE. However, the extent of LGE was not different in patients with and without cardiac events. Moreover, the AS pattern of LGE was the best independent predictor of the combined endpoint. The AS group had higher troponin release than other groups but lower levels of inflammatory markers such as ESR and CPR. This result suggests a different kind of myocarditis, with less inflammation but greater myocardial damage.
These findings were concordant with those of a previous postmortem study by Shirani et al. (19) and with a report by Mahrholdt et al. (11), who found similar distributions of LGE patterns in a nonselected population of 86 patients with AM. Mahrholdt et al. (11) found that the IL pattern of LGE was mostly associated with the detection of the parvovirus B19 genome, whereas the AS pattern was associated with the human herpesvirus 6 genome and with the combined presence of both the parvovirus B19 and herpesvirus 6 genomes. Moreover, in that study, the presence of LGE in the ventricular septum was the strongest CMR predictor of chronic ventricular dysfunction, as well as ventricular dilatation at follow-up. In accord with the results of Mahrholdt et al. (11), we found that the AS pattern was most prevalently associated with worsening of LVEF at follow-up than other LGE presentations.
The difference in LGE pattern could be explained by the different tropism of viruses. In fact, human herpesvirus 6 infects not only T cells but also cells of the nervous system and the cardiac conduction system, and it can establish a latent state after primary infection (20). After a primary infection in early childhood, AM may be a reactivation of the disease, with myocardial infection and damage occurring in the septum because of the presence of virus in the cardiac conduction system (21). Repeated viral reactivation may be the cause of the most frequent progression to LV dysfunction found in patients with the septal pattern of LGE than in those without it. Moreover, reactivation of viral infection in the cardiac conduction system may be the trigger for arrhythmic events in patients with the AS pattern. In contrast, parvovirus B19 is associated with polyserositis and pericarditis after initial viremia (22). Thus the left lateral free wall and the inferior wall may be involved because of the direct contact with the pericardium.
We did not perform endomyocardial biopsy in all the patients in the study because our population was composed of hemodynamically stable patients with preserved LVEF, and the invasive procedure was not indicated. Thus we cannot evaluate the presence of different viral genomes in our patients, and we did not know whether these or other viruses were involved in the mechanism of damage of our patients. However, our results strengthen the role of AS LGE in myocarditis by the finding of a prognostic role of this pattern during long-term follow-up of patients with preserved LVEF. Future studies are needed to evaluate whether the different patterns of LGE and different prognoses could be caused by different viral tropisms or other factors.
First, as mentioned, we did not perform endomyocardial biopsy in all the patients, and the diagnosis was made by the summation of clinical and CMR findings. As shown in the 2009 white paper in the Journal, CMR criteria are highly specific for the diagnosis of myocarditis but less sensitive (5). Our population is almost completely composed of patients with infarct-like myocarditis, as demonstrated by the high prevalence of chest pain (95%), new ECG abnormalities (96%), elevations in CRP and/or ERS (99%), and troponin increases (100%). As a previous study demonstrated, in patients with infarct-like presentations, CMR is also very sensitive for detection of AM (17). In contrast, CMR has a low sensitivity in patients with heart failure presentations and a very low sensitivity in those with arrhythmic presentations, and for this reason we decided to exclude patients with these presentations. Furthermore, when compared with endomyocardial biopsy, CMR was demonstrated to be very accurate for the detection signs of myocardial damage in AM, whereas it was less sensitive for the diagnosis of chronic myocarditis (23). In our population of patients with clinically suspected AM and an infarct-like presentation, CMR may be considered both specific and sensitive to confirm the diagnosis of AM.
Second, we did not perform T1 and T2 mapping in our population. T1 mapping could permit the detection and quantification of microscopic fibrosis, even in the absence of positive LGE, whereas T2 mapping would allow us to detect myocardial edema quantitatively. However, at the time of patients’ enrollment in the study, these 2 techniques were not available in all CMR scanners, and nowadays, the pulse sequence of T1 and T2 mapping may provide different results for different CMR vendors. Therefore, in the context of a multicenter study, we preferred not to include these 2 new yet promising techniques.
In this multicenter study, patients with AM and preserved LVEF had different patterns of LGE. The AS pattern of LGE was associated with a worse prognosis and with presentations different from those the IL pattern, the no-LGE pattern, and other patterns.
COMPETENCY IN MEDICAL KNOWLEDGE: CMR enabled confirmation of AM in hemodynamically stable patients with clinically suspected AM and preserved EF. In these patients the 2 most frequent patterns of LGE were the subepicardial IL and the AS midwall pattern. Patients presenting with LGE in the midwall of the AS myocardial segments had a worse prognosis than did patients with other patterns of distribution.
TRANSLATIONAL OUTLOOK: Further research should be conducted to determine whether the different patterns of presentation of LGE in AM are associated with different viral causes.
Dr. Pontone has received institutional grants and fees from GE Healthcare, Medtronic, Bracco, Bayer, and Heartflow. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute myocarditis
- cardiac magnetic resonance
- C-reactive protein
- end-diastolic volume
- end-diastolic volume index
- ejection fraction
- erythrocyte sedimentation rate
- implantable cardioverter-defibrillator
- late gadolinium enhancement
- left ventricular
- right ventricular
- steady-state free precession
- Received May 11, 2017.
- Revision received July 19, 2017.
- Accepted August 8, 2017.
- 2017 American College of Cardiology Foundation
- Caforio A.L.,
- Pankuweit S.,
- Arbustini E.,
- et al.
- Kindermann I.,
- Barth C.,
- Mahfoud F.,
- et al.
- Friedrich M.G.,
- Sechtem U.,
- Schulz-Menger J.,
- et al.
- Grün S.,
- Schumm J.,
- Greulich S.,
- et al.
- Cheong B.Y.,
- Muthupillai R.,
- Wilson J.M.,
- et al.
- Wong T.C.,
- Piehler K.M.,
- Zareba K.M.,
- et al.
- Mahrholdt H.,
- Goedecke C.,
- Wagner A.,
- et al.
- Mahrholdt H.,
- Wagner A.,
- Deluigi C.C.,
- et al.
- Abdel-Aty H.,
- Boye P.,
- Zagrosek A.,
- et al.
- Perfetti M.,
- Malatesta G.,
- Alvarez I.,
- et al.
- Hundley W.G.,
- Bluemke D.A.,
- Finn J.P.,
- et al.
- Di Bella G.,
- Siciliano V.,
- Aquaro G.D.,
- et al.
- Aquaro G.D.,
- Camastra G.,
- Monti L.,
- et al.
- Francone M.,
- Chimenti C.,
- Galea N.,
- et al.
- Lurz P.,
- Eitel I.,
- Adam J.,
- et al.