Author + information
- Received June 25, 2018
- Revision received August 15, 2018
- Accepted August 16, 2018
- Published online October 17, 2018.
- Snigdha Jain, MDa,
- Daniel Kuriakoseb,
- Ilaina Edelsteinb,
- Bilal Ansari, MBBSb,
- Garrett Oldland, MDc,
- Swetha Gaddam, MDb,c,
- Khuzaima Javaid, MDc,
- Pritika Manaktala, MDb,
- Jonathan Leeb,d,
- Rachana Miller, MDc,d,
- Scott R. Akers, MD, PhDc and
- Julio A. Chirinos, MD, PhDb,c,d,∗ ()
- aDepartment of Internal Medicine, UT Southwestern Medical Center, Dallas, Texas
- bDepartment of Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
- cDepartments of Internal Medicine and Radiology, Corporal Michael J. Crescenz VAMC, Philadelphia, Philadelphia
- dDepartment of Medicine, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania
- ↵∗Address for correspondence:
Dr. Julio A. Chirinos, South Tower, Room 11-138, Perelman Center for Advanced Medicine, 3400 Civic Center Boulevard, Philadelphia, PA 19104.
Objectives This study researched right atrial (RA) deformation indexes and their association with all-cause mortality among subjects with or without heart failure (HF).
Background Although left atrial dysfunction is well described in HF, patterns of RA dysfunction and their prognostic implications are unclear. Cardiac magnetic resonance (CMR) imaging can provide excellent visualization of the RA. We used CMR to characterize RA phasic function in HF and to assess its prognostic implications.
Methods This study prospectively examined 608 adults without HF (n = 407), as well as adults with HF with a reduced ejection fraction (HFrEF) (n = 105) or with HF with a preserved ejection fraction (HFpEF) (n = 96). Phasic RA function was measured via volume measurements and feature-tracking methods to derive longitudinal strain. All-cause death was ascertained over a median follow-up of 38.9 months. Standardized hazard ratios (HRs) were computed via Cox regression.
Results Measures of RA phasic function were more prominently impaired in subjects with HFrEF than those in subjects with HFpEF. In analyses that adjusted for demographic factors, HF status, left ventricular ejection fraction, right ventricular end-diastolic volume index, and right ventricular ejection fraction, RA reservoir strain (HR: 0.66; 95% confidence interval [CI]: 0.47 to 0.92; p = 0.0154), RA expansion index (HR: 0.53; 95% CI: 0.31 to 0.91; p = 0.0116), RA conduit strain (HR: 0.58; 95% CI: 0.40 to 0.84; p = 0.0039), and RA conduit strain rate (HR: 1.51; 95% CI: 1.02 to 2.220; p = 0.0373) independently predicted all-cause mortality. In contrast, RA booster pump function and RA volume index did not independently predict the risk of death.
Conclusions Phasic RA function is predictive of the risk of all-cause death in a diverse group of subjects with and without HF. RA conduit and reservoir function are independent predictors of mortality.
Left atrial (LA) enlargement (1–3) and phasic function (3–10) are known to be associated with worse cardiovascular morbidity and mortality in various clinical conditions, particularly in hypertension and congestive heart failure (HF). However, the presence of right atrial (RA) dysfunction in HF, and its prognostic implications, have not been comprehensively studied.
The RA is now being recognized as more than a passive filling chamber. During the cardiac cycle, it serves 3 primary functions: 1) a reservoir function, in which it acts as a reservoir for the systemic venous return during ventricular systole and atrial diastole; 2) a conduit function, in which it allows blood to flow passively into the ventricle during early ventricular diastole; and 3) a booster pump function, by which it actively pumps blood into the right ventricle (RV) in late ventricular diastole (11,12). Despite its physiological importance, patterns of RA dysfunction in HF with preserved ejection fraction (HFpEF) and reduced ejection fraction (HFrEF) have not been characterized. Furthermore, data regarding the prognostic implications of RA remodeling are scarce.
In this study, we aimed to: 1) compare indexes of phasic RA function (based on phasic volumes and longitudinal RA strain measured with cardiac magnetic resonance imaging [CMR]) among subjects with HFpEF, HFrEF, and in subjects without HF; 2) assess the prognostic value (mortality risk) of various measures of RA phasic function; and 3) assess whether measures of phasic RA function are independently predictive of the risk of death, even after adjustment for demographic, clinical, and left ventricular (LV) parameters.
We enrolled a convenience clinical sample of 608 subjects who underwent a clinically indicated outpatient CMR study. Indications for CMR included assessment of LV or LV and/or right ventricular (RV) function (27.6%), assessment of myocardial ischemia (22.7%), assessment of the presence and/or pattern of delayed enhancement (9.9%), simultaneous assessment of aortic dimensions and cardiac function (7.9%), severe diastolic dysfunction (8.22%), evaluation for ruling out cardiac thrombus (5.9%), assessment of regional wall motion (2.6%), syncope (0.2%), myocardial viability (2.6%), quantification of mitral regurgitation (1.32%), or miscellaneous indications (11%). The protocol was approved by the Philadelphia VA Medical Center Institutional Review Board, and written informed consent was obtained from all participants.
HFrEF was defined as symptomatic HF in the presence of left ventricular ejection fraction (LVEF) <50%. HFpEF was defined as: 1) New York Heart Association functional class II to IV symptoms consistent with HF, in the absence of significant aortic stenosis; 2) LVEF >50%; and 3) a mitral E wave to annular e′ ratio >14 (13), or at least 2 of the following: a) mitral E wave to annular e′ ratio >8; b) treatment with a loop diuretic for control of HF symptoms; c) LA volume index >34 ml/m2 of body surface area; d) NT−proB-type natriuretic peptide level >200 pg/ml; and e) LV mass index >149 g/m2 in men and 122 g/m2 in women (measured by CMR). Subjects without HF had an LVEF >50% without symptoms and signs consistent with HF.
Key exclusion criteria were as follows: 1) claustrophobia; 2) presence of metallic objects or implanted medical devices in body; 3) atrial fibrillation, flutter, or significant arrhythmia at the time of enrollment, which might have compromised the study measurements; and 4) other conditions that would make the study measurements less accurate or unreliable (i.e., inability to perform an adequate breath hold for CMR acquisitions, poor gating that prevented a clear visualization of the RA wall, poor image quality). The latter condition excluded 54 subjects. At baseline, a CMR study was performed, which was used for volumetric and RA strain analyses. All-cause mortality was ascertained over a median period of 38.9 months.
CMR imaging protocol
Participants underwent a CMR using a 1.5-T whole body magnetic resonance image scanner (Avanto or Espree, Siemens, Malvern, Pennsylvania) equipped with a phase-array cardiac coil. LV and RV volumes, and EF were determined using balanced steady-state, free-precession cine imaging. Typical parameters were as follows: repetition time: 30.6 ms; echo time: 1.3 ms; phases: 30; slice thickness: 8 mm; matrix size: 192 × 192; and integrated parallel imaging technique factor: 2. The endocardial borders were manually traced in end-systole and end-diastole in a short-axis cine stack to quantify LV and RV volumes, and EF. LV mass was computed as the difference between epicardial and endocardial volumes in end-diastole, multiplied by myocardial density. LV mass was normalized for body height in meters raised to the allometric power of 1.7 (14).
Assessment of RA function
RA analyses were performed using CVI42 image analysis software (Circle Cardiovascular Imaging Inc., Calgary, Ontario, Canada) and QRS-gated 4-chamber steady state free precision cine acquisitions. RA endocardial and epicardial borders were manually traced in the 4-chamber view using LV end-diastole as the phase of reference, as well as manual specification of the long axis of the RA. This was followed by application of an automated tracking algorithm based on 2-dimensional cross-correlation that followed the motion of image landmarks. Strain was computed as the average wall deformation relative to the longitudinal axis of the RA. Tracking performance was visually reviewed to ensure accurate tracking of the atrial myocardium. In case of insufficient automated border tracking, manual adjustments were made to the initial contour, and the algorithm was reapplied. An example of atrial wall tracking is shown in Figure 1 and in Online Video 1.
Values of segmental deformation were exported and further processed using custom software programmed in Python (Python Software Foundation, Wilmington, Delaware). We computed longitudinal atrial strain, which was defined as the change of atrial myocardial length throughout the atrial cycles (L1) compared with its resting (or reference) length (L0) in a relaxed state at diastasis (end of atrial diastole), as (L1 − L0)/L0. We noted that this method differed from computation of RA strain using ventricular diastole as the reference length. We computed reservoir strain as the sum of the absolute values of conduit and booster pump strain (Figure 1). Coefficients of variation assessed with consecutive repeated measurements (including de novo plane prescription) in 4 volunteers were: 11.1%, 11.1%, and 10.1% for RA conduit, reservoir, and booster strain, respectively; 5.9%, 2.3%, and 7.7% for maximum, minimum, and diastolic volume, respectively; 11.5%, 12.4%, and 15.3% for the RA expansion index, passive, and active emptying fractions, respectively; and 23% and 11.8% for RA conduit and booster pump strain rates, respectively.
We first compared general clinical characteristics of patients without HF, HFrEF, and HFpEF. We used chi-square tests for categorical variables and analysis of variance with post hoc pairwise comparisons with Bonferroni correction for continuous variables. We also compared measures of RA strain among the groups. The prognostic value of various parameters was assessed using Cox regression, to obtain hazard ratios (HRs) and 95% confidence intervals (CIs). Statistical significance was defined as a 2-tailed p value <0.05. All p values presented are 2-tailed. Statistical analyses were performed using the Matlab statistics and machine learning toolbox (Mathworks, Natwick, Massachusetts) and SPSS for Windows version 22 (IBM, Armonk, New York).
We included 407 subjects without HF, 105 subjects with HFrEF, and 96 subjects with HFpEF. Baseline characteristics of study participants stratified by HF status (HFrEF, HFpEF, and no HF) are presented in Table 1. Subjects with HFpEF and HFrEF were slightly older than subjects without HF. All groups were predominantly men, although the proportion of women was greater in subjects with HFpEF compared with those with HFrEF. The study sample was largely composed of white and African-American participants. Body mass index, and the prevalence of obstructive sleep apnea and hypertension were higher in the HFpEF group, whereas coronary artery disease was more common in the HFrEF group. Levels of N-terminal pro–B-type natriuretic peptide and low-density lipoprotein cholesterol were higher in subjects with HFrEF compared with those with HFpEF and those without HF.
Comparison of RA function among groups
A comparison of RA function indexes among the 3 groups is shown in Table 2 and Figures 2 and 3. There were highly significant differences in indexes of RA function among the groups. In post hoc pairwise comparisons, subjects with HFrEF demonstrated a reduced RA reservoir strain and RA expansion index (measures of reservoir function), reduced conduit strain and passive emptying fraction (measures of conduit function), and a reduced active emptying fraction (measure of booster pump function) compared with subjects with HFpEF and those without HF. Passive emptying fraction, an index of conduit function, was also reduced in subjects with HFpEF compared with those without HF.
After adjusting for age, sex, race/ethnicity, body mass index, systolic blood pressure, history of chronic obstructive pulmonary disease, obstructive sleep apnea, current smoking, diabetes mellitus, RV end-diastolic volume index, RVEF, and pulmonary artery systolic pressure (PASP) (Table 2, Figure 3), between-group differences persisted across measures of RA reservoir and conduit function, but not booster function, when subjects with HFrEF were compared with those without HF. In these adjusted analyses, compared with subjects without HF, those with HFrEF had a reduced RA reservoir strain and expansion index (reflecting reservoir function) and a reduced RA conduit strain and passive emptying fraction (reflecting conduit function). In these adjusted analyses, subjects with HFpEF demonstrated a reduced RA expansion index (reflecting reservoir function) compared with those without HF.
Table 2 and Figure 3 demonstrate comparisons of RA phasic volumes (maximum RA volume index, minimum RA volume index, and diastatic RA volume index) among the groups. The HFrEF group demonstrated greater minimum and diastatic RA volume than that in subjects without HF. After adjusting for age, sex, race/ethnicity, body mass index, systolic blood pressure, history of chronic obstructive pulmonary disease, obstructive sleep apnea, current smoking, diabetes mellitus, RV end-diastolic volume index, RVEF, and PASP, the differences between subjects with HFrEF and those without HF persisted for the RA minimum volume index but not the diastatic volume index. In subjects with HFpEF, RA volumes were not significantly different from those without HF.
Prognostic value of RA function
Over a median duration of 38.9 months, 78 subjects died. Table 3 and Figures 4 and 5 demonstrate the results of proportional hazards (Cox) models in which RA strain measures were assessed as predictors of all-cause death. All HRs were standardized to facilitate an intuitive comparison of the association between different indexes of RA structure and function, with the risk of death.
In unadjusted analyses (Table 3, Figure 4), both volumetric and functional indexes of the RA reservoir and conduit function were significant predictors of death. The strongest associations were seen for measures of RA reservoir function: RA expansion index (HR: 0.49; 95% CI: 0.33 to 0.75; p = 0.0009) and RA reservoir longitudinal strain (HR: 0.70; 95% CI: 0.54 to 0.91; p = 0.007). Measures of conduit RA function (RA passive emptying fraction, RA conduit longitudinal strain and strain rate) were also significantly predictive of mortality (HR values >1.00 for conduit strain rate and <1 for reservoir and conduit strains indicated that greater reductions in reservoir and conduit function were associated with a greater risk of death). Although the RA active emptying fraction was significantly predictive of mortality (HR: 0.72; 95% CI: 0.57 to 0.90; p = 0.0049), booster pump strain and booster pump strain rate were not significantly predictive.
The areas under the receiver-operating characteristic curves for unadjusted Cox models that included the RA reservoir strain and RA expansion index were 0.603 and 0.629, respectively. The areas under the receiver-operating characteristic curve for unadjusted Cox models that included RA conduit strain, the conduit strain rate, and passive emptying fraction were 0.606, 0.583, and 0.594, respectively. The area under the receiver-operating characteristic curve for an unadjusted Cox model that included the active emptying fraction was 0.598.
In analyses that adjusted for age, sex, race, body mass index, systolic blood pressure, diabetes, chronic obstructive pulmonary disease, obstructive sleep apnea, current smoking, LVEF, HF status, RV end-diastolic volume index, RVEF, and PASP (Table 3, Figure 5), measures of reservoir function–RA reservoir strain (HR: 0.66; 95% CI: 0.47 to 0.92; p = 0.0154) and RA expansion index (HR: 0.53; 95% CI: 0.31 to 0.91; p = 0.0224), as well as measures of conduit function - RA conduit strain (HR: 0.58; 95% CI: 0.40 to 0.84; p = 0.0039) and RA conduit strain rate (HR: 1.51; 95% CI: 1.02 to 2.22; p = 0.0373) remained significantly predictive of the risk of death. In adjusted analyses, measures of booster function (RA booster pump strain and active emptying fraction) were not significant predictors of mortality.
The areas under the receiver-operating characteristic curves for multivariable Cox models that included RA reservoir strain and the expansion index were 0.738 and 0.740, respectively. The areas under the receiver-operating characteristic curves for multivariable Cox models that included conduit strain and the conduit strain rate were 0.737 and 0.739, respectively.
Prognostic value of RA volumes
In unadjusted analyses, all measures of phasic RA volumes (RA maximum volume index, RA minimum volume index, and RA diastatic volume index) significantly predicted the risk of death. However, after adjustment for age, sex, race, LVEF, HF status, RV end-diastolic volume index, and RVEF, none of the volume indexes was independently predictive of death.
In this prospective study, we comprehensively compared patterns of phasic RA function among subjects with HFrEF, HFpEF, and those without HF, and assessed the association between measures of RA function and mortality. Applying novel feature-tracking algorithms to CMR cine images, we demonstrated abnormalities in reservoir and conduit RA function (RA reservoir strain, expansion index, conduit strain, and passive emptying fraction) in HFrEF and reservoir function in HFpEF (RA expansion index) compared with subjects without HF. Furthermore, we reported that measures of RA reservoir and conduit function, but not booster pump function, were significant predictors of mortality, even after adjusting for multiple confounders, including LVEF, HF status, RV end-diastolic volume index, and RVEF. Our findings indicated that RA remodeling and dysfunction represented an important cardiac phenotype in patients with or at risk of HF.
RA phasic function in HFpEF versus HFrEF
RA phasic function in HF is largely unknown, and to our knowledge, this was the first study that studied differences in RA function in patients with HFrEF and HFpEF compared with subjects without HF. We comprehensively assessed RA phasic function using both volumetric measures (RA expansion index, passive RA emptying fraction, and active RA emptying fraction) and strain-based measures (reservoir strain, conduit strain and strain rate, booster strain and strain rate) based on feature-tracking cine CMR.
We observed that RA minimum volume was higher in those with HFrEF compared with subjects without HF. Volumetric indexes of reservoir (RA expansion index) and conduit (RA passive emptying fraction) function, but not booster function, were reduced in subjects with HFrEF compared with those without HF. In addition, longitudinal strain-based measures of RA reservoir and conduit function were reduced, with no significant differences in booster longitudinal strain. In contrast, only the RA expansion index, a measure of RA reservoir function, was significantly reduced in subjects HFpEF compared with subjects without HF in adjusted analyses.
CMR imaging offers an excellent opportunity to assess atrial volumes and phasic deformation, and thereby, to study atrial function (15). Previous studies that evaluated RA function in HF are sparse. In a single study that compared RA function in patients with HFrEF with those with no HF using echocardiography, the RA peak systolic strain and the peak systolic strain rate, which represent reservoir function, were impaired, whereas RA early diastolic and late diastolic strain (reflecting conduit and booster pump function, respectively) were preserved (16). In humans with pulmonary arterial hypertension, using 2-dimensional speckle-tracking echocardiography, RA reservoir and conduit function were noted to be impaired, although booster function was preserved (17). A similar pattern with a reduced RA passive emptying fraction (reflecting conduit function) but increased active emptying fraction (reflecting booster pump function) was described in healthy subjects with increasing age, which suggested that reduced diastolic RV function with age might lead to reduced conduit and improved booster pump function as a compensatory response (18). These studies collectively suggested, that even in the presence of frank disease, booster pump function tends to be preserved. Accordingly, in our study, we found that reservoir and conduit function were reduced in HFrEF, whereas booster pump function was not significantly different when compared with subjects without HF.
Prognostic value of RA function
We found that RA reservoir strain and the expansion index (measures of RA reservoir function), and RA conduit strain and the conduit strain rate (measures of RA conduit function) were associated with an increased risk of death, independent of demographic characteristics, HF status, and CMR-derived LVEF, LV mass, RV end-diastolic volume index, RVEF, and PASP.
The role of right heart function is being increasingly recognized in the assessment of patients with HF (12). RV dysfunction has been shown to be of prognostic value in patients with both HFpEF and HFrEF (19,20). Because RV dysfunction can be a late consequence of changes in the pulmonary circulation that result from HF, changes in RA dimension and phasic function can aid in the assessment of right heart function. Because RA phasic function may respond to changes in RV compliance and diastolic function early in the course of HF, it could potentially identify a phenotype to better risk stratify patients.
In previous studies, the maximum RA volume index (RAVI) was shown to be associated with adverse outcomes. In patients with chronic systolic HF, RAVI, as assessed by echocardiography, was predictive of a composite endpoint of death, transplantation, and/or hospitalization for HF (21). This association was reproduced in another study that used CMR to examine RA dimensions, which found RAVI to be an independent predictor of mortality in patients with HFrEF (22). In our study, we studied a diverse group of participants, including subjects with HFrEF, HFpEF, and without HF. In addition to the assessment of RA volume, we performed a comprehensive assessment of RA phasic function using feature-tracking algorithms and compared it among subjects with HFpEF and HFrEF, and those without HF. Furthermore, this study was the first to examine the prognostic value of phasic RA function using CMR in subjects without HF, HFpEF, and HFrEF. We found that the RA reservoir and conduit strain, the RA expansion index and RA conduit strain, and conduit strain rate were independent predictors of mortality. In contrast to our study, Sato et al. (23) found that the RA reservoir volume index, as assessed by CMR, was predictive of clinical worsening (defined as hospitalization because of right HF, lung transplantation, or pulmonary hypertension−related death) in patients with pulmonary hypertension. After adjusting for demographic factors, HF status and LVEF and RVEF, we did not find any of the volume indexes to be independent predictors of mortality.
Our study should be interpreted in the context of its strengths and limitations. The strengths of our study were the inclusion of a large clinical sample in a prospective cohort design; comprehensive assessment of atrial phasic function using both volume-based methods and indexes of phase function derived from phasic longitudinal strain that were measured using novel tissue-tracking techniques; use of CMR, which is the gold standard technique for assessment of LVEF, LV mass, and volumes; inclusion of subjects with both HFpEF and HFrEF in comparison to those without HF; and adjusted statistical analyses that accounted for gold standard magnetic resonance imaging measures of LV mass and LVEF, RV size, and function. This study also had some limitations. This study included a convenience clinical sample that might have limited the generalizability of our results to unselected clinical or community-based cohorts. The predominantly male population in our study sample limited the generalizability of our results to women; future studies should assess the prognostic value in populations with greater sex diversity and determine whether differences exist in this regard. We did not measure a 3-dimensional RA volume, but computed it based on a single plane using the Simpson’s rule. However, previous data demonstrated that RA volume assessed by this method were strongly correlated with 3-dimensional RA volume measured from transverse axial plane multislice 3-dimensional reconstructions (24).
In conclusion, our study was the first to report a comprehensive evaluation of patterns of RA dysfunction in HFpEF and HFrEF, and to study its prognostic value in predicting all-cause mortality. We demonstrated that RA reservoir and conduit function are impaired in subjects with HFrEF compared with those without HF, whereas only the RA expansion index was reduced in HFpEF in adjusted analyses. In addition, we found that measures of RA reservoir and conduit function were independent predictors of death, in a diverse group of subjects with HFrEF, HFpEF, and without HF, independent of various demographic and clinical factors, as well as LV mass, LVEF, RV size, and RVEF.
COMPETENCY IN MEDICAL KNOWLEDGE: RA function can be comprehensively assessed using CMR. RA function is impaired in HFrEF. Impaired RA reservoir and conduit function independently predict all-cause death.
TRANSLATIONAL OUTLOOK: Standard cine imaging and novel feature tracking techniques offer the opportunity to comprehensively assess the RA using CMR imaging. RA function is a prognostic phenotype that should be studied in more detail using noninvasive imaging. Further studies should assess whether RA function can also identify patients with HF who might preferentially benefit from therapies that target the pulmonary circulation.
Dr. Chirinos was supported National Institutes of Health grants R56HL-124073-01A1, R01-HL-121510-01A1, and 5-R21-AG-043802-02, and a VISN-4 research grant from the Department of Veterans Affairs; has received consulting honoraria from Bristol-Myers Squibb, OPKO Healthcare, Fukuda Denshi, Microsoft, Ironwood Pharmaceuticals, Sanifit, Pfizer, Merck, and Bayer; has received research grants from the American College of Radiology Network, Fukuda Denshi, Bristol-Myers Squibb, Microsoft; and has been named as inventor in a University of Pennsylvania patent application for the use of inorganic nitrates and/or nitrites for the treatment of heart failure and preserved ejection fraction. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- cardiac magnetic resonance
- heart failure
- heart failure with preserved ejection fraction
- heart failure with reduced ejection fraction
- hazard ratio
- left atrium
- left ventricle
- left ventricular ejection fraction
- pulmonary artery systolic pressure
- right atrium
- right atrial volume index
- right ventricle
- Received June 25, 2018.
- Revision received August 15, 2018.
- Accepted August 16, 2018.
- 2018 American College of Cardiology Foundation
- Benjamin E.J.,
- Dagostino R.B.,
- Belanger A.J.,
- Wolf P.A.,
- Levy D.
- Appleton C.P.,
- Kovacs S.J.
- Hoit B.D.
- Kurt M.,
- Wang J.,
- Torre-Amione G.,
- Nagueh S.F.
- Santos A.B.,
- Roca G.Q.,
- Claggett B.,
- et al.
- Sugimoto T.,
- Bandera F.,
- Generati G.,
- Alfonzetti E.,
- Bussadori C.,
- Guazzi M.
- Bhave N.M.,
- Visovatti S.H.,
- Kulick B.,
- Kolias T.J.,
- McLaughlin V.V.
- Gorter T.M.,
- van Veldhuisen D.J.,
- Bauersachs J.,
- et al.
- Maceira A.M.,
- Cosin-Sales J.,
- Prasad S.K.,
- Pennell D.J.
- Querejeta Roca G.,
- Campbell P.,
- Claggett B.,
- Solomon S.D.,
- Shah A.M.
- Guazzi M.,
- Dixon D.,
- Labate V.,
- et al.
- Ghio S.,
- Gavazzi A.,
- Campana C.,
- et al.
- Sallach J.A.,
- Tang W.H.,
- Borowski A.G.,
- et al.
- Ivanov A.,
- Mohamed A.,
- Asfour A.,
- et al.
- Sato T.,
- Tsujino I.,
- Ohira H.,
- et al.
- Ebtia M.,
- Murphy D.,
- Gin K.,
- et al.