Author + information
- Received July 24, 2014
- Revision received December 12, 2014
- Accepted December 15, 2014
- Published online March 17, 2015.
- Yousef Etoom, MD∗,
- Sindu Govindapillai, MD∗,
- Robert Hamilton, MD∗,
- Cedric Manlhiot, BSc∗,
- Shi-Joon Yoo, MD∗,†,
- Maryam Farhan, MD∗,
- Samir Sarikouch, MD‡,
- Brigitte Peters, MD§,
- Brian W. McCrindle, MD∗ and
- Lars Grosse-Wortmann, MD∗,†∗ ()
- ∗The Labatt Family Heart Centre, Department of Paediatrics, The Hospital for Sick Children, University of Toronto, Toronto, Canada
- †Department of Diagnostic Imaging, The Hospital for Sick Children, University of Toronto, Toronto, Canada
- ‡Department of Heart, Thoracic, Transplantation and Vascular Surgery, Hannover Medical School, Hannover, Germany
- §Institute for Biometry and Medical Informatics, University of Magdeburg, Magdeburg, Germany
- ↵∗Reprint requests and correspondence:
Dr. Lars Grosse-Wortmann, The Labatt Family Heart Centre, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada.
Background Cardiac magnetic resonance (CMR) is a component of the revised Task Force Criteria (rTFC) for the diagnosis of arrhythmogenic right ventricular cardiomyopathy (ARVC). However, its diagnostic value in a pediatric population is unknown.
Objectives This study examined the contribution of CMR to diagnosing ARVC using the rTFC in a pediatric population.
Methods Clinical CMR studies of 142 pediatric patients evaluated for ARVC between 2005 and 2009 were reviewed. Patients were categorized into “definitive,” “borderline,” “possible,” or “no” ARVC diagnostic groups based on the rTFC. The extent to which each element of the rTFC contributed to diagnosing ARVC was determined using a c-statistics model.
Results A total of 23 (16%), 32 (23%), 37 (26%), and 50 (35%) patients had definite, borderline, possible, and no ARVC, respectively, applying the rTFC. The prevalence of regional wall motion abnormalities in these groups was 83%, 53%, 22%, and 16%, respectively (p < 0.001). By CMR, right ventricular end-diastolic volumes were 118 ± 31 cc/m2, 108 ± 22 cc/m2, 94 ± 14 cc/m2, and 92 ± 18 cc/m2, respectively (p < 0.001). Right ventricular fatty infiltration and fibrosis were detected in only 1 and 3 patients, respectively, all of whom had definitive ARVC. Of all rTFC major criteria, CMR had the largest c-statistic decline (c = −0.163). Eleven of the 23 patients (48%) with definite ARVC would not have been in this group if CMR had not been performed.
Conclusions CMR parameters are important contributors to a diagnosis of ARVC in children, using the rTFC. Fatty infiltration and myocardial fibrosis provide limited value in children and adolescents.
An inherited cardiomyopathy with an autosomal dominant pattern, arrhythmogenic right ventricular cardiomyopathy (ARVC) is characterized by progressive fibrofatty replacement of the right ventricular (RV) myocardium (1–3). Clinically, the disease is distinguished by ventricular arrhythmias and right or biventricular systolic dysfunction (4). Although ARVC is rare, it is thought to be responsible for 20% or more of sudden cardiac deaths among young individuals (5,6).
In the absence of a true gold standard, the 2010 revision of the 1994 Task Force Criteria (rTFC) (Table 1), developed by a multinational task force and published simultaneously in Europe and the United States, is used to diagnose ARVC in today’s practice (7). The rTFC are a composite score of structural, histological, electrical, and genetic features of the disease, yielding a diagnostic certainty of either “no,” “possible,” “borderline,” or “definite” ARVC. The most important difference between the original criteria and the rTFC lies within a quantitative and more defined assessment of RV size and function (7).
Even with the rTFC, diagnosing ARVC in pediatric patients remains challenging. In addition to the recognized shortcomings of the rTFC (8–10), children who are gene-positive for ARVC may not exhibit all of the disease’s phenotypic features until adulthood. Cardiac magnetic resonance (CMR) evaluation to assess RV end-diastolic volume index (EDVi), ejection fraction (EF), and presence of wall motion abnormalities (WMAs) is part of the rTFC scoring system and has been shown to be a useful noninvasive tool to diagnose ARVC in adults (11). CMR has also been advocated for the detection of fibrofatty degeneration of myocardium in adults (12–14).
Despite a postulated advantage of CMR in detecting early stages of ARVC (15), its usefulness in diagnosing ARVC in the pediatric population has yet to be determined. The aims of this study were to: 1) evaluate the prevalence of CMR findings in children diagnosed with ARVC; and 2) assess the importance of CMR and other components of the rTFC for the diagnosis of ARVC.
Following approval by the institutional research ethics board, all CMR studies performed between 2005 and 2009 in pediatric patients referred for first-time evaluation of suspected ARVC at the authors’ institution were reanalyzed. At the time of referral, patients also underwent echocardiography, electrocardiogram (ECG), signal-averaged ECG, and 24-h Holter monitoring. Each diagnostic modality was analyzed in a single laboratory at the authors’ institution, following uniform protocols and reading criteria, although not by a single reader. Echocardiographic studies included parasternal long- and short-axis views to assess regional WMAs and right ventricular outflow tract (RVOT) diameter as well as an apical 4-chamber view for RV fractional area change. Where available, results of endomyocardial biopsy were recorded. As all endomyocardial biopsies were clinically graded prior to the revision of the Task Force Criteria in 2010, it was not possible to differentiate between major and minor histological criteria. Consequently, the presence of qualitative fibrofatty infiltration as per the original Task Force Criteria was included as a positive major criterion for the purpose of this study (7).
CMR scans were performed on a 1.5-T scanner (Signa CV/I, General Electric Medical Systems, Milwaukee, Wisconsin; or Avanto, Siemens Medical Solutions, Erlangen, Germany). The CMR protocol included cine imaging in axial, short-axis, and 2-chamber planes using the steady-state free precession technique; double inversion fast spin-echo acquisitions for myocardial fat imaging; and late gadolinium enhancement for the detection of myocardial fibrosis. The cine short-axis stack was analyzed in the routine clinical fashion to determine RV volumetric parameters using commercially-available software (Mass Analysis, Medis Medical Imaging Systems, Leiden, the Netherlands).
All CMR studies were retrospectively reviewed for presence and severity of WMAs, fatty infiltration, and myocardial fibrosis by an experienced reader (L.G-W.) who was blinded to the results of other diagnostic tests, including ventricular volumes, as well as to the overall classification according to the rTFC. The locations of fatty infiltration, fibrosis, and WMAs were described using the RV 9-segment model, as described by Wald et al. (16). Additional qualitative CMR findings in the RV included the presence of aneurysms, wall thinning, accordion sign, outflow tract dilation, abnormal trabeculations, and hypertrophy.
Patients with concomitant congenital heart disease or no/insufficient axial or short-axis cine imaging were excluded. Further, patients lacking clinical data in more than 2 of the rTFC criteria categories were excluded.
Demographic information and test results of the study population are presented either as means and SD or as frequencies, where appropriate. Differences in prevalence of CMR findings among the 4 categories of ARVC diagnostic certainty were assessed by linear or logistic regression using ARVC diagnosis as a categorical variable with no ARVC as the reference category. A p value <0.05 was considered statistically significant. To compare EDVi and EF across ages and sexes, the values were converted into standard deviation scores (SDS) as per the lambda-mu-sigma method by relating individual values for ventricular volumes and EFs to their respective 50th percentiles (SDS = 0) in age- and sex-specific healthy control subjects (17). The interpretation of SDS is similar to that of z scores. In contrast to z scores, however, SDS generation does not rely on normally distributed data and is regarded as more accurate, particularly when relatively small samples are examined. A detailed description of this approach can be found elsewhere (18). To determine the contribution of each rTFC major criterion to the diagnosis of ARVC, the c-statistic was used. For the c-statistic model to be applicable, the 4 ARVC diagnostic categories were reduced to a dichotomous variable: “yes” or “no” ARVC. Definite and borderline diagnostic groups were considered to have ARVC, whereas possible and no groups were considered to not have ARVC. Each major diagnostic criterion was entered into the regression as a dichotomous (yes/no) variable. The model was run multiple times, each time withholding a single diagnostic criterion. The relative contribution-to-model fit of each rTFC major criterion was determined by the difference between the c-statistic for the overall model and the c-statistic for the model without that particular criterion. This approach is similar to a previously published report on adult ARVC (3). The statistical significance of the difference between the area under the curve obtained from the different models was assessed using the DeLong method (19). The primary reader reassessed a randomly selected sample of 20 CMR studies after an interval of at least 6 weeks following the primary reading, and a second reader (S.J.Y.) assessed the same sample for interobserver variability. Interrater and intrarater reliability were assessed using kappa statistics. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, North Carolina).
CMR studies from 213 patients referred for first-time imaging evaluation for ARVC were reviewed. Reasons for referral included family history in 100 patients (47%), cardiac symptoms in 81 (38%), ventricular arrhythmias in 75 (35%), incidentally discovered abnormal ECG findings in 4 (2%), and incidental abnormal echocardiographic findings in 6 (3%). Twenty-eight CMR studies were excluded for incomplete or poor quality short-axis or axial cine imaging, 23 for concomitant congenital heart disease, and 20 because the rTFC criteria in 3 or more categories were unavailable. Demographics and prevalence of rTFC findings in the remaining 142 study subjects are shown in Tables 2 and 3⇓⇓. Eighty patients (56%) were male and 62 (44%) were female. Their mean age was 13.8 ± 3.2 years, with no significant difference between diagnostic groups (p = 0.35). Of the 142 included patients, 50 (35%) had no ARVC, 37 (26%) possible ARVC, 32 (23%) borderline ARVC, and 23 (16%) definite ARVC. The distribution of the 4 categories did not differ when patients were stratified by reasons for referral (Table 2). Thirty-five (26%) subjects had undergone endomyocardial biopsy, with a positive biopsy in 6 of 9 (67%) patients with definite ARVC, 3 of 11 (27%) with borderline ARVC, and no patients with possible or no ARVC. One-third of the study population (44 subjects) had undergone genetic testing for known pathogenic ARVC mutations. Of those with genetic testing results, a pathogenic mutation was found in 4 of 9 (44%) patients with definitive, 3 of 13 (23%) with borderline, 1 of 20 (5%) with possible, and 0 of 7 with no ARVC.
Regional WMAs were present in 19 of 23 (83%) patients with definite ARVC, 17 of 32 (53%) with borderline, 8 of 37 (22%) with possible, and 8 of 50 (16%) with no ARVC (p < 0.001) (Table 4). Of 52 patients with WMAs, 47 (90%) had disturbances in the proximal RVOT (segments 2, 5, and/or 8, according to the classification by Wald et al. ), 47 (71%) in the distal RVOT (segments 1, 4, and/or 7), and 26 (50%) in the remote RV free wall (3, 6, and/or 9). A total of 40 of the 52 patients with WMAs had at least 2 affected segments, and 21 had WMAs in 3 or more segments. RV EDVi was significantly different among ARVC diagnostic groups (p < 0.001), with larger volumes in the groups with higher diagnostic certainty of ARVC. RV EF was lower in subjects with a definitive diagnosis of ARVC compared with the other diagnostic groups, although this difference failed to reach statistical significance (p = 0.07). The RV EDVi SDS increased gradually and significantly with progression from no to definite ARVC (p < 0.001) (Table 4).
RV fatty infiltration and myocardial fibrosis were detected in 1 and 3 patients, respectively, all of whom had definite ARVC per the rTFC. Grossly abnormal RV trabeculations, accordion sign, and RV wall thinning varied significantly across diagnostic groups (p = 0.002, p < 0.001, and p = 0.04, respectively) and were more prevalent in the diagnostic groups with greater ARVC certainty (Table 5). Patients with definite ARVC had significantly larger left ventricles (LVs) than those with borderline, possible, or no ARVC (LV EDVi 97.2 ± 19.1 cc/m2 vs. 95.0 ± 19.9 cc/m2, 87.3 ± 11.8 cc/m2, and 85.4 ± 16.3 cc/m2, respectively; p < 0.001). LV EDVi SDS correlated with RV EDVi SDS (r = 0.63; p < 0.0001), and LV EF SDS correlated with RV EF SDS (r = 0.36; p < 0.0001).
Intrarater and interrater variability for all qualitative CMR parameters is summarized in Table 6. The agreement for intrarater and interrater reliability was good for RVOT dilation, RV trabeculation, accordion sign, and RV wall thinning, and was moderate for presence of WMAs (17,20).
Contribution of CMR findings to the overall performance of the rTFC
A c-statistics model was used to evaluate the contribution of each major criterion within the rTFC to the performance of the overall score. Minor criteria were not evaluated, either because they were not assessed (biopsies), had a low prevalence (repolarization abnormalities), were, relatively speaking, evenly distributed across the 4 diagnostic certainty groups (this was the case for CMR, family history, and arrhythmia), or because of low specificity (as for depolarization abnormalities and echocardiographic abnormalities). To differentiate between contributions from CMR and echocardiography, these 2 tests were considered as separate variables in this model. In this cohort, only 2 patients met major criteria for echocardiography and none met minor criteria.
With all major criteria included in the logistic regression model, a c-statistic of 0.933 was achieved (Central Illustration). The rTFC major criteria ranked as follows, in order of descending c-statistic decline, indicating decreasing contribution of the individual criterion for the performance of the rTFC: abnormal CMR (c = −0.163 ± 0.038; p < 0.001), positive family history (c = −0.121 ± 0.031; p < 0.001), biopsy (c = −0.038 ± 0.020; p = 0.06), abnormal depolarization (c = −0.008 ± 0.009; p = 0.25), echocardiographic findings (c = −0.002 ± 0.024; p = 1.00), abnormal repolarization (no change, c = 0.000 ± 0.024; p = 1.00), and presence of arrhythmia (c = +0.002 ± 0.024; p = 1.00). In summary, echocardiography, repolarization abnormalities, and arrhythmia did not make significant contributions to the diagnosis of ARVC in this population. In contrast, CMR, family history, biopsy, and depolarization abnormalities fully accounted for the performance of the diagnostic model, with a combined c-statistic of 0.933 ± 0.024, which was similar to the performance of a diagnostic model including all parameters (p = 1.00). Almost one-half (48%) of the patients with definitive ARVC (11 of 23) would have been in a group with lower diagnostic certainty without taking into account their CMR findings.
In the absence of a single conclusive test, the diagnosis of ARVC in children remains challenging, especially early in the clinical course when structural, functional, and electrical alterations in the RV may be minimal. CMR has become a powerful diagnostic imaging modality for anatomical delineation, functional analysis, and tissue characterization in congenital and acquired heart disease, including adult ARVC (12–14,21).
The prevalence of CMR findings in adult patients with suspected ARVC has been described both in the era of the original criteria and of the rTFC (10,22). In contrast, limited information is available on the prevalence of CMR findings in children (23,24). Only 1 previous publication systematically described the prevalence of CMR abnormalities in children referred for ARVC assessment, but it was beyond that study’s scope to stratify subjects by whether or not they had ARVC (25). To the best of our knowledge, this is the first study to detail the prevalence as well as the extent of WMAs, RV dilation, and systolic dysfunction in a large pediatric population with ARVC status elucidated by rTFC.
In adults with this condition, the RV is significantly larger compared with control subjects (7,26,27). Because RV enlargement is part of the rTFC, this association between RV size and diagnostic certainty is to be expected; our results corroborate this observation in a pediatric cohort. After conversion of RV EDVi values to SDS to eliminate sex- and age-related fluctuations, a disproportionate RV enlargement in the groups with higher diagnostic certainty persisted.
Interestingly, whereas RV EF remained stable across the first 3 diagnostic groups, RV EDVi (absolute and SDS) increased gradually with higher diagnostic certainty for ARVC. Although this is a cross-sectional study and any longitudinal conclusions from it remain speculative, the discrepancy between earlier RV enlargement and relatively late deterioration in EF may suggest that pediatric ARVC patients have relatively preserved global RV systolic function during early stages of the disease, whereas RV enlargement begins early in the course.
Further, all 4 ARVC diagnostic groups contained individuals with abnormal RV EF by SDS, including the unaffected ARVC group. There are several possible reasons for this observation. Cutoffs for normal RV EF may not be applicable to pediatric populations. Alternatively, the unaffected and possible ARVC groups in this study may contain patients who are not entirely “normal” and may develop more signs of the condition at an older age. More importantly for the context of this study, however, is the fact that as ARVC diagnostic certainty increases, the RV EF SDS becomes more abnormal (Table 4).
Although regional WMAs are commonly found in patients with ARVC (26–28), little is known about exactly where in the RV these disturbances occur in ARVC. In our cohort, most WMAs were located within the RV outflow tract, a common origin of electrical ventricular ectopy in ARVC. Whether this finding represents a true predilection for this site or reflects easier perception on imaging is unclear. The high prevalence of WMAs, as well as the performance of CMR in the c-statistics model, suggests that recognition of WMAs is important in securing an ARVC diagnosis. However, WMAs can be difficult to assess in clinical routine as reflected by only moderate intrarater and interrater reproducibility for WMAs in the current study. New methods to quantify regional RV contractility by CMR in a less user-dependent fashion may prove useful in detecting and quantifying RV WMAs more objectively and reproducibly (29).
Role of CMR in diagnosing ARVC in children
The current study suggests that CMR contributes importantly to the diagnosis of ARVC in the setting of early disease in a pediatric age group: CMR was “needed” to achieve a diagnostic certainty of definite in one-half of all patients in that category. Of all major criteria within the rTFC, CMR made the greatest contribution to the performance of the diagnostic score. There was poor agreement of CMR findings (and overall score) with echocardiographic findings of RV dysfunction and enlargement, with only 2 patients in the entire cohort meeting major echocardiographic rTFC criteria and none meeting minor criteria. In many conditions, echocardiography and CMR correlate well, but in the presence of pathology the agreement between the 2 modalities is typically weaker than in healthy control subjects (18). Furthermore, the 2 modalities tend to correlate best for continuous variables (20,30). The rTFC are based on categorical variables (presence or absence of WMAs) or dichotomized continuous variables (cutoffs for RVOT diameter or fractional area change). In this situation, agreement between echocardiography and CMR becomes much less reliable, especially when different measurements are compared (RVOT diameter and fractional area change by echocardiography vs. RV volume and RV EF by CMR).
A study by Marcus et al. (3) revealed echocardiography as the most important criterion in the framework of the original Task Force Criteria for adults. In their study, CMR contributed only marginally to the overall score. This discrepancy in comparison with our results may stem from 2 differences between the 2 studies. First, Marcus et al. (3) used the original Task Force Criteria; in the current study, participants were evaluated according to the revised TFC. Second, RV regional dysfunction and dilation are likely more subtle in children and adolescents than in adults. Echocardiography may be insensitive to these early changes of ARVC due to limitations in acoustic windows to the RV free wall as compared with CMR. Consequently, as anticipated by Basso et al. (15) in their 2009 review paper, CMR may prove to be a more suitable tool than echocardiography in the detection of early functional and anatomical signs of the condition. In contrast, the abnormalities in adults may be sufficiently pronounced to be detected via echocardiography.
Another reason why CMR has been advocated in ARVC is that it has a unique ability to noninvasively detect fibrosis and fatty infiltration of the RV in patients with suspected ARVC, although fat and fibrosis detection by noninvasive imaging is not part of the rTFC (21,31). However, in clinical practice, the unequivocal detection of fatty infiltration of the RV by CMR is often challenging (32). In our cohort, only 1 patient had convincing evidence of fatty infiltration. This patient, along with 2 others, also had fibrosis of the RV myocardium. All 3 had definite ARVC and also had other imaging findings, including WMAs, dilated RVs, and reduced RV EF. Given the paucity of convincing fibrofatty infiltration in our cohort and the fact that it was always associated with other imaging abnormalities, we believe that fat-sensitive sequences and late gadolinium enhancement carry low incremental yields in pediatric patients with suspected ARVC.
Besides fibrofatty infiltration of the myocardium, other structural changes have been identified in patients with ARVC, including focal RV wall thinning, abnormal RV trabecular architecture, and the so-called “accordion sign” (33). In our cohort, all of these abnormalities demonstrated reasonable reproducibility on intrarater assessment (κ between 0.669 and 0.775) and inter-rater assessment (κ between 0.706 and 0.801) and differed significantly across groups of diagnostic certainty. The magnitude of the kappa coefficient for each qualitative CMR parameter was lower than expected, likely given the small sample size. Nonetheless, these abnormalities should be reported if present.
Traditionally, and reflected by its name, ARVC has been regarded as a disease primarily of the RV. More recently, Asimaki et al. (34) elegantly showed that the same desmosomal changes that cause myocyte loss and electrical instability in the RV are also present in the interventricular septum and in the LV. On a macroscopic level, regional LV dysfunction is not uncommon in patients with ARVC (35). Interestingly, patients in our cohort who had definite or borderline ARVC had larger LVs than those with possible or no ARVC. Further, RV enlargement and dysfunction also was associated with LV enlargement and dysfunction. Whether LV changes in these patients occur independently of RV pathology or result from ventricular interaction in the presence of predominantly right-sided disease remains to be clarified.
Interestingly, a mutation associated with ARVC was identified in less than one-half of the patients with definite ARVC who underwent genetic testing. The most likely explanation is that there are yet undetected or untested genes responsible for ARVC diagnosis or that patients are so labeled but, in reality, have a different condition.
The absence of an independent diagnostic gold standard made it impossible to calculate specificity and sensitivity for the individual components of the rTFC. Instead, like Marcus et al. (3) before us, we used a c-statistics model to assess the diagnostic performance of each rTFC parameter. Nevertheless, the fact that the ultimate diagnostic group of no to definite ARVC depends on the variables whose diagnostic merits are being evaluated remains a strong confounder. Although this limits our current study, it cannot be avoided and is at the root of an imperfect clinical diagnostic tool to this day (3). There was a paucity of patients in our study cohort who underwent endomyocardial biopsy. In most centers, including ours, children and adolescents undergo biopsy more selectively than adults. Last, although the rTFC do not include “hypokinesia” as a positive WMA criterion, we believe that akinesia and hypokinesia represent a spectrum of regional disturbances of contractility. The distinction can be difficult and is, in fact, arbitrary. Given that most children are expected to be in early stages of ARVC, a decision was made to include hypokinesia as a positive WMA finding before analysis was begun.
CMR findings are frequent in children and adolescents with a suspicion for ARVC. CMR parameters listed in the rTFC, including RV EDVi and WMAs, are important contributors to diagnosing ARVC in this age group. Although widely measured, fatty infiltration and myocardial fibrosis rarely occur in children and offer little incremental value during a CMR examination. LV enlargement and systolic dysfunction, on the other hand, are common and can provide important clues to the diagnosis. Our study suggests that CMR is more useful as a diagnostic imaging modality than echocardiography for ARVC diagnosis in the context of the rTFC.
COMPETENCY IN PATIENT CARE: The diagnostic evaluation of pediatric patients with suspected ARVC generally relies upon the European Society of Cardiology’s 2010 rTFC, which address the structural, histological, electrical, and genetic features of the disease. The family medical history and results of CMR imaging are particularly important factors to consider.
TRANSLATIONAL OUTLOOK: Further studies are needed to elucidate the diagnostic and prognostic significance of LV involvement in patients with signs and symptoms of ARVC, particularly as these evolve though childhood and adolescence.
The study was approved by the Research Ethics Board of the Hospital for Sick Children, Toronto, Ontario, Canada. The study was performed at the Hospital for Sick Children. This work was supported in part through the University of Toronto ‘Comprehensive Research Experience for Medical Students (CREMS)’ Program and by the German Competence Network for Congenital Heart Defects, funded by the German Federal Ministry of Education and Research (BMBF) (FKZ 01G10210, 01GI0601). All authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Etoom and Govindapillai contributed equally to this work.
- Abbreviations and Acronyms
- arrhythmogenic right ventricular cardiomyopathy
- cardiac magnetic resonance
- end-diastolic volume index
- ejection fraction
- left ventricle/ventricular
- right ventricle/ventricular
- revised Task Force Criteria
- standard deviation scores
- wall motion abnormality
- Received July 24, 2014.
- Revision received December 12, 2014.
- Accepted December 15, 2014.
- American College of Cardiology Foundation
- Marcus F.I.,
- Fontaine G.H.,
- Guiraudon G.,
- et al.
- Corrado D.,
- Basso C.,
- Thiene G.
- Dalal D.,
- Nasir K.,
- Bomma C.,
- et al.
- Marcus F.I.,
- McKenna W.J.,
- Sherrill D.,
- et al.
- Gomes J.,
- Finlay M.,
- Ahmed A.K.,
- et al.
- Wald R.M.,
- Haber I.,
- Wald R.,
- Valente A.M.,
- Powell A.J.,
- Geva T.
- Sarikouch S.,
- Peters B.,
- Gutberlet M.,
- et al.
- Sarikouch S.,
- Koerperich H.,
- Dubowy K.O.,
- et al.
- Tandri H.,
- Castillo E.,
- Ferrari V.A.,
- et al.
- Sen-Chowdhry S.,
- Prasad S.K.,
- Syrris P.,
- et al.
- Tandri H.,
- Saranathan M.,
- Rodriguez E.R.,
- et al.
- Groenink M.,
- Wilde A.A.
- Jain A.,
- Shehata M.L.,
- Stuber M.,
- et al.