Author + information
- Received October 22, 2015
- Revision received January 6, 2016
- Accepted January 12, 2016
- Published online March 29, 2016.
- Jong-Hwa Ahn, MDa,
- Sung Mok Kim, MDb,
- Sung-Ji Park, MD, PhDa,∗ (, )
- Dong Seop Jeong, MD, PhDc,
- Min-Ah Woo, MSd,
- Sin-Ho Jung, PhDd,
- Sang-Chol Lee, MD, PhDa,
- Seung Woo Park, MD, PhDa,
- Yeon Hyeon Choe, MD, PhDb,
- Pyo Won Park, MD, PhDc and
- Jae K. Oh, MDa,e
- aDivision of Cardiology, Department of Medicine, Cardiovascular Imaging Center, Heart Vascular Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea
- bDepartment of Radiology, Cardiovascular Imaging Center, Heart Vascular Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea
- cDepartment of Thoracic Surgery, Heart Vascular Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea
- dBiostatistics and Clinical Epidemiology Center, Research Institute for Future Medicine, Samsung Medical Center, Seoul, Republic of Korea
- eDivision of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota
- ↵∗Reprint requests and correspondence:
Dr. Sung-Ji Park, Division of Cardiology, Department of Internal Medicine, Cardiovascular Imaging Center, Heart Vascular Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81, Irwon-ro, Gangnam-gu, Seoul 135-710, Republic of Korea.
Background Although a common symptom in patients with severe aortic stenosis (AS) without obstructive coronary artery disease (CAD), little is known about the pathogenesis of exertional angina.
Objectives This study sought to prove that microvascular dysfunction is responsible for chest pain in patients with severe AS and normal epicardial coronary arteries using adenosine-stress cardiac magnetic resonance (CMR) imaging.
Methods Between June 2012 and April 2015, 117 patients with severe AS without obstructive CAD and 20 normal controls were enrolled prospectively. After exclusions, study patients were divided into 2 groups according to presence of exertional chest pain: an angina group (n = 43) and an asymptomatic group (n = 41), and the semiquantitative myocardial perfusion reserve index (MPRI) was calculated.
Results MPRI values were significantly lower in severe AS patients than in normal controls (0.90 ± 0.31 vs. 1.25 ± 0.21; p < 0.001), and were much lower in the angina group than the asymptomatic group (0.74 ± 0.25 vs. 1.08 ± 0.28; p < 0.001). In logistic regression analysis, the only independent predictor for angina was MPRI (odds ratio: 0.003; p < 0.001). Univariate associations with MPRI were identified for diastolic blood pressure, E/e′ ratio, left ventricular volume and ejection fraction, cardiac index, presence of late gadolinium enhancement, and left ventricular mass index (LVMI). In multivariate analysis, LVMI was the strongest contributing factor to MPRI (standardization coefficient: -0.428; p < 0.001).
Conclusions Our results suggest that, in patients with severe AS without obstructive CAD, angina is related to impaired coronary microvascular function along with LV hypertrophy detectable by semiquantitative MPRI using adenosine-stress CMR. Clinical Trial Registration: NCT02575768
The onset of symptoms marks a dramatic decline in the prognosis of aortic stenosis (AS) (1); thus, it is the most important indication for aortic valve replacement (AVR) (2). Among its symptoms, angina is 1 of the 3 most common and occurs frequently in the absence of epicardial coronary artery disease (CAD) (3–5). Previous studies have demonstrated that angina in patients with normal coronary arteries can be attributed to left ventricular (LV) hypertrophy (LVH), which can cause coronary ischemia due to increased LV oxygen demand and impaired myocardial perfusion reserve (MPR) (4,6). This phenomenon has been revealed by invasive coronary catheterization (7) and imaging modalities, such as positron emission tomography (8) and cardiac magnetic resonance (CMR) (9,10). In the absence of significant coronary stenosis, coronary ischemia is indicative of microvascular dysfunction, but it remains unsettled whether the reduced MPR seen in severe AS without obstructive CAD leads to chest pain in response to stress stimuli.
Adenosine-stress CMR can detect stress-induced abnormal hypoperfusion with signs and symptoms of ischemia without CAD (11,12), and is a reliable noninvasive imaging method that allows assessment of the transmyocardial distribution of coronary blood flow and the MPR index (MPRI). We hypothesized that microvascular dysfunction is responsible for chest pain in patients with severe AS and nonobstructive epicardial coronary arteries. Therefore, adenosine-stress CMR was performed in patients with severe AS and normal epicardial coronary arteries to compare semiquantitative MPRI between patients with and without chest pain. Additionally, MPRI of patients with severe AS was compared with that of normal controls.
Patients with severe AS were enrolled prospectively from a single tertiary care center, Samsung Medical Center in South Korea, between June 2012 and April 2015. Subjects who had severe AS and preserved LV ejection fraction (LVEF), defined as ≥50% when assessed by transthoracic echocardiography, were included in this study (Figure 1). Severe AS was defined as an indexed aortic valve area (AVA) of <0.6 cm2/m2 per guidelines (13). Patients with any of the following criteria were excluded: age <18 years; other concomitant valvular disease of at least moderate severity; previous AVR; obstructive epicardial CAD (>30% luminal stenosis in at least 1 coronary artery on coronary angiography); history of myocardial infarction or acute coronary syndrome; contraindication to adenosine; any absolute contraindication to CMR; or estimated glomerular filtration rate of <30 ml/min/1.73 m2. In total, 117 severe AS patients without obstructive CAD and with preserved LVEF were screened. We excluded 33 patients with predominant symptoms other than exertional chest pain, such as dyspnea on exertion, syncope, or mixed symptoms. The remaining 84 patients became the subjects of this study and were divided into an asymptomatic group and an angina group based on their predominant presenting symptom at baseline. Twenty healthy asymptomatic subjects with no history of cardiovascular disease, diabetes, or hypertension (mean age: 65.3 years; males: 40%) were recruited to serve as normal controls. The Institutional Review Board of Samsung Medical Center approved this study and all subjects gave written informed consent before the investigation.
Dyspnea on exertion was defined as greater than New York Heart Association functional class II. Angina was defined as exertional chest pain. Those with chest heaviness or chest discomfort were included in the angina group. Those with dizziness or presyncope were included in the syncope group. To classify patient symptoms, all initial symptoms assessed by the primary physician in the medical records were carefully reviewed by 1 cardiologist (S.J.P).
Coronary angiograms were analyzed quantitatively at the angiographic core laboratory (Heart Center, Samsung Medical Center, Seoul, South Korea) with an automated edge-detection system (Centricity CA 1000; GE, Waukesha, Wisconsin) using standard definitions (14). An independent interventional cardiologist interpreted all coronary angiograms. We included only patients without obstructive CAD (<30% reduction in 3 major epicardial arteries and the largest first order branches of each major epicardial artery).
Comprehensive transthoracic echocardiography (M-mode, 2-dimensional, and Doppler) was performed with a dedicated unit (Vivid 7; GE Healthcare, Port Washington, New York). All measurements were performed in accordance with the current American Society of Echocardiography and European Association of Echocardiography guidelines (15,16). The mean transaortic pressure gradient and peak transaortic velocity (Vmax of AV) were measured in all available views and the highest values were used for analysis. The time–velocity integral at the aortic valve and LV outflow tract levels were acquired through continuous wave and pulse wave Doppler echocardiography, respectively, and the AVA was calculated with the continuity equation (17). The average of 3 consecutive Doppler signals was used.
All patients underwent CMR in a 1.5-T scanner (Magnetom Avanto, Syngo MR B17 version; Siemens Medical Solutions, Erlangen, Germany) with a 32-channel phased-array receiver coil. After localization, cine images of the LV were acquired through a steady-state free-precession sequence in short axis, 4-chamber, and 2-chamber views. For stress MR perfusion imaging, adenosine was injected intravenously at 0.14 mg/kg/min for 3 min, both before and during imaging. The perfusion stress sequence was performed with intravenous infusion of gadobutrol at 0.1 mmol/kg body weight at an injection rate of 3 ml/s, followed by a 30-ml saline flush. Ten minutes after stress perfusion imaging, a second bolus of 0.1 mmol/kg gadobutrol was injected and rest perfusion images were obtained. Spoiled gradient echo techniques (turboFLASH pulse sequences) were used for the stress and rest perfusion sequences. Eighty dynamic CMR images were acquired at each slice location with a 1 min to 1 min 30 s total acquisition time depending on heart rate.
Data were analyzed offline by means of a commercially available dedicated software tool (Dynamic Signal Analysis, Argus, Siemens Medical Solutions) at our CMR core laboratory by 1 experienced cardiologist (J.H.A) and 1 experienced radiologist (S.M.K), who were blinded to the clinical information of the patients. For functional analysis, LV contours were drawn manually on the short-axis images at the end of diastole and of systole, and the contours were propagated automatically to other images for the rest of the cardiac cycle. Then, manual correction of the automatically rendered endocardial and epicardial contours was performed for all datasets, and papillary muscles as well as myocardial trabeculations were included in the ventricular cavity. On the basis of these data, the following parameters were calculated: LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LVEF, LV mass, stroke volume (SV), right ventricular end-diastolic volume (RVEDV), right ventricular end-systolic volume (RVESV), and right ventricular ejection fraction (RVEF). Cardiac output was calculated as the product of SV and heart rate. The LV mass index (LVMI), SV index, and cardiac index were calculated as LV mass, SV, and cardiac output divided by the body surface area. Late gadolinium enhancement (LGE) images for focal fibrosis, categorized as present or absent, were qualitatively assessed.
First-pass perfusion images at stress and rest states were analyzed semiquantitatively. Subendocardial and subepicardial borders were positioned on each slice on a frame with high contrast between the LV cavity and the myocardium. The borders were propagated automatically onto all frames. An interactive correction was done if necessary. Care was taken to exclude any linear dark rim artifacts at the LV cavity/endocardial border. The LV cavity region of interest was adjusted manually so that it would include the region of maximal signal intensity within the cavity and exclude papillary muscle. An American Heart Association 16-segment model was used (true apex not imaged), resulting in a total of 32 segments per patient (16 for the stress examination and 16 for the resting examination). Signal intensity–time curves were generated for all segments and the maximum upslope of the LV myocardium divided by the maximum upslope of the LV cavity. The MPRI [upslopestress(corrected)/upsloperest(corrected)] was calculated as the ratio of the segmental upslope values during adenosine and rest, as previously described (18,19). Whole (average of all myocardial segments) MPRIs were calculated (Figure 2). The interobserver agreement of 2 different experienced observers was assessed on different days for 20 randomly selected patients.
Statistical analysis was performed with SAS for Window, version 9.4 (SAS Institute Inc., Cary, North Carolina). Continuous variables are expressed as mean ± SD and categorical variables are presented as absolute numbers and proportions (%). Overall comparisons between groups were performed with Student t test for continuous variables and the chi-square or Fisher exact test when the Cochran rule was not met for categorical variables. E/e′ ratio was log-transformed to improve the normality of its transformation. The presence of angina was regressed on age, sex, body mass index, systolic blood pressure, diastolic blood pressure, left atrial volume index, early diastolic mitral inflow (E) velocity, late diastolic mitral inflow (A) velocity, tissue Doppler-derived early diastolic mitral annular (e′) velocity, Vmax of AV, mean pressure gradient, AVA, right ventricular systolic pressure, E/e′ ratio, LVEDV, LVESV, LVEF, RVEDV, RVESV, RVEF, SV, cardiac index, LVMI, and presence of LGE using the logistic regression method. To minimize the impact of missing data points in the regression analysis: 1) missing data points were imputed by their sample means (there was no missing in categorical variables); 2) a backward variable selection was applied to the imputed data; and 3) multivariable logistic regression analysis using only the selected variables as covariates was applied to the cases with complete data points for the selected variables. Relationships between MPRI and variables were assessed in univariate linear regression analysis and multivariate linear regression analysis. MPRI was regressed on the same set of covariates using the same variable selection procedure and the standard multiple linear regression method. A 2-sided p < 0.05 was considered significant.
Additional study method details are in the Online Appendix.
Clinical and echocardiographic characteristics
Of the 84 AS patients without obstructive CAD, 43 (51%) were in the angina group and 41 (49%) were in the asymptomatic group (Figure 1). No differences were observed between the 2 groups in their demographic characteristics or underlying diseases. Baseline clinical characteristics of the study patients and normal controls are shown in Table 1.
According to baseline echocardiographic parameters of the angina and asymptomatic groups (Table 2), severity of AS, as determined by the AVA, AVA index, peak velocity of AV, and mean pressure gradient, did not differ significantly between the 2 groups. Diastolic parameters (left atrial volume index, E/A ratio, e′, and E/e′ ratio), LVEDV, LVESV, and LVEF also did not differ significantly between the 2 groups.
Adenosine stress CMR results
Every normal control had normal cardiac structure, LVEF, RVEF, SV index, and cardiac index on CMR. Patients with AS had significantly smaller RVEDV, RVESV, LVEF, SV index, and cardiac index values and larger LVESV, RVEF, and LVMI values than normal controls. Between the 2 groups with severe AS, there were no differences in CMR parameters including presence of LGE, except for the LVMI and MPRI values (Table 3). The LVMI was significantly higher in the angina group than in the asymptomatic group (109.08 ± 37.88 vs. 93.72 ± 101.58; p = 0.032). The mean MPRI was significantly lower in patients with severe AS than in normal controls (0.90 ± 0.31 vs. 1.25 ± 0.21; p < 0.001); in particular, the angina group had a much lower value than the asymptomatic group (0.74 ± 0.25 vs. 1.08 ± 0.28; p < 0.001). To examine whether lower MPRI was an independent predictor of angina, a multivariate analysis was performed using clinical, echocardiographic, and adenosine-stress CMR parameters. Logistic regression analysis revealed that only MPRI remained as an independent predictor of angina (Table 4). Representative adenosine-stress CMR images and MPRI values for the angina and asymptomatic groups are shown in Figure 3 and the Online Appendix. The mean MPRIs for the normal controls and the 2 groups of severe AS patients without obstructive CAD are shown in Figure 4.
Factors associated with MPRI and interobserver agreement
Univariate and multivariate linear regression analyses were performed to identify clinical, echocardiographic, and adenosine-stress CMR determinants associated with the MPRI (Table 5). Univariate analyses showed that the MPRI correlated with diastolic blood pressure, e′, and LVEF and was inversely related to E/e′, LVEDV, LVESV, cardiac index, LVMI, and presence of LGE. On multivariate analysis, LVMI and LGE were independently associated with MPRI. A relationship between MPRI and LVMI is shown in Figure 5.
Excellent interobserver agreement was found for the semiquantitative MPRI (intraclass correlation coefficient = 0.934). The mean of the differences between the 2 observers was -0.046 and the 95% limits of agreement were -0.352 to 0.260. A Bland–Altman result for the interobserver agreement of the MPRI measurements is displayed in Online Figure 1.
The principal findings of this study are that the semiquantitative MPRI was significantly lower in patients with severe AS without obstructive CAD than in normal controls; and among the patients with severe AS, MPRI was much lower in the angina group than in the asymptomatic group. The clinical and echocardiographic parameters (including AS severity) and CMR parameters, other than the MPRI, did not differ between patients with and without angina consistent with our previous study (20). Therefore, our results suggested that coronary microvascular dysfunction is a responsible mechanism for developing chest pain in patients with severe AS without obstructive CAD (Central Illustration). Additionally, LVMI was the single strongest determinant for MPRI, a finding consistent with a previous study (9).
Pathophysiology of symptomatic presentation of severe AS
The cardinal manifestations of AS include syncope, angina, and dyspnea. It has been well-described that patient survival becomes limited once they develop AS symptoms and depends on the types of symptoms that develop (21). In our previous study (20), we reported why specific symptoms develop and whether the severity of AS (within the range of severe AS) or the cardiac hemodynamic response to AS determined the occurrence of specific symptoms. In patients with severe AS who developed dyspnea, a markedly altered LV diastolic function with increased filling pressure occurred. Conversely, patients who developed syncope presented with decreased SV, LV, and left atrial values. However, we could not determine the mechanisms of development of chest pain in our previous study. Therefore, this study sought to assess the pathophysiology of this development in patients with severe AS and normal epicardial coronary arteries.
Prior invasive coronary catheterization studies have demonstrated that coronary flow reserve is lower in AS subjects than in normal subjects (4,7); in particular, 1 study indicated that patients with angina had significantly lower coronary flow reserves than those without angina (4). A noninvasive and precise measurement of microvascular function can be performed through quantification of myocardial blood flow using adenosine-stress CMR. Our adenosine-stress CMR results were consistent with much of the data in the prior 2 studies and reaffirmed the proposed hypothesis that reduced coronary flow reserve could cause angina in AS patients without obstructive CAD. The concordance of these abnormal findings, obtained by independent methods, strongly supports the notion that reduced myocardial flow reserve is a key pathogenic component of angina without obstructive CAD.
Mechanisms of coronary microvascular dysfunction in hypertrophic hearts
Although the development of LVH in patients with AS is an adaptive and compensatory response, pathologic hypertrophy can lead to coronary microvascular dysfunction despite the presence of angiographically normal coronary arteries. Pathogenetic mechanisms of coronary microvascular dysfunction include structural changes (vascular rarefaction and perivascular fibrosis) and functional changes (endothelial dysfunction and dysfunction of smooth muscle) (22,23). Functional and structural changes of the coronary microcirculation have been well-documented in all models of pathologic LVH (23,24). Furthermore, in patients with LVH induced by AS, extravascular alterations also contributed to impairment of microvascular function. Extravascular compression and increased diastolic perfusion time have been proposed as the main mechanisms for improvement in myocardial blood flow and coronary flow reserve after AVR (25). A recent study also showed that coronary diastolic suction is reduced with increasing heart rate in patients with severe AS and appropriate increase in coronary diastolic suction with increasing heart rate was restored immediately after transcatheter AVR (26).
The role of adenosine-stress CMR in severe AS
Adenosine-stress CMR is a validated noninvasive method to detect obstructive CAD and demonstrate abnormal coronary flow reserve (19,27,28). Adenosine-stress CMR also has been used to demonstrate microvascular dysfunction based on a reduced MPR in patients with syndrome X (11,12), nonischemic dilated cardiomyopathy (29), systemic lupus erythematosus with chest pain (30), and repaired coarctation of the aorta without obstructive CAD (31). In patients with AS, 2 previous studies showed that severe AS patients had lower MPRs than normal controls and that these reserves decreased significantly with increasing NYHA functional class (10,19). However, these 2 studies did not examine the differences in the MPRI values of AS patients with and without angina.
CMR can be useful in assessing patients with AS in several aspects. First, because CMR provides a dimensionally accurate and 3-dimensional perspective of the heart, it has been considered by many to be the gold standard for measurements of ejection fraction, volumes, and cardiac index. Second, AVA by CMR correlates well with transesophageal echocardiography planimetry in patients with AS (32), and CMR velocity-encoded imaging can accurately measure the antegrade velocity through the stenotic aortic valve without angle dependence (33). Thus, CMR can be used to assess the severity of the aortic valve and AVA. Third, myocardial fibrosis detected by CMR in patients with AS has potential prognostic value. The extent of LGE correlated well with fibrosis on myocardial biopsy and also correlated inversely with the extent of symptom improvement and survival after AVR (34,35). Finally, impairment of MPR, as assessed by adenosine-stress CMR, has been closely associated with decreased exercise capacity and severity of dyspnea (9). Now, we have shown that the impairment of MPR is associated with the presence of angina in severe AS without obstructive CAD. Thus, MPRI has the potential to classify and clarify symptom status in patients with severe AS who had ambiguous symptoms.
First, the study group comprised a relatively small number of patients enrolled in a single center. Before adenosine-stress CMR can be used routinely in clinical practice, further large prospective studies are required. However, this study included the largest number of patients with severe AS without obstructive CAD for adenosine-stress CMR of any comparable investigation to date. Second, we included a considerable number of AS patients with diabetes and hypertension, both of which might contribute to myocardial perfusion abnormalities (36). However, previous studies have shown a consistent decrease in MPRI in severe AS patients (9,10), with no significant difference in the prevalence of diabetes and hypertension in the angina and asymptomatic groups in our study. Our results are, therefore, likely to be widely generalizable. Third, patients with LV systolic dysfunction, which could be inversely related to MPRI (29), were not included in our study because such patients may have hemodynamic changes related to LV systolic dysfunction. Therefore, our study results are not applicable to patients with LV systolic dysfunction. Finally, we did not evaluate the reversibility of MPRI after intervention nor the natural history or clinical outcome of asymptomatic patients with decreased MPRI value. Therefore, we could not assess the usefulness of MPRI as a biomarker. Currently, several studies are ongoing (37), including ours evaluating the reversibility and prognostic value of MPRI; hence, a clinically valuable biomarker role of MPRI could be established in the future.
Compared with severe AS patients without any symptoms, severe AS patients with angina but no obstructive CAD demonstrated a reduced MPR, which is indicative of microvascular dysfunction. Furthermore, a reduced MPR was associated with LVH. Our results suggested that angina in patients with severe AS without obstructive CAD can been attributed to LVH, which can cause myocardial ischemia by coronary microvascular dysfunction.
COMPETENCY IN MEDICAL KNOWLEDGE: Exertional angina in patients with severe AS without obstructive CAD can be attributed to LVH, which can cause myocardial ischemia by coronary microvascular dysfunction.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: The assessment of the MPR through adenosine-stress CMR can confirm the presence of angina and may predict the imminent onset of angina in patients with severe AS without obstructive CAD. Thus, MPRI shows promise as a potential classifier or clarifier of symptoms in patients with severe AS who had ambiguous symptoms.
TRANSLATIONAL OUTLOOK: As this study did not evaluate reversibility of the MPRI after AVR and prognosis of asymptomatic patient with decrease MPRI, additional research is needed to understand the reversibility of coronary microvascular dysfunction and prognostic value of MPRI as a biomarker.
For an expanded Methods section as well as a supplemental figure and videos, please see the online version of this article.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Ahn and Kim contributed equally to this work.
- Abbreviations and Acronyms
- aortic stenosis
- aortic valve area
- aortic valve replacement
- coronary artery disease
- cardiac magnetic resonance
- late gadolinium enhancement
- left ventricular
- left ventricular end-diastolic volume
- left ventricular ejection fraction
- left ventricular end-systolic volume
- left ventricular hypertrophy
- left ventricular mass index
- myocardial perfusion reserve
- myocardial perfusion reserve index
- right ventricular end-diastolic volume
- right ventricular ejection fraction
- right ventricular end-systolic volume
- stroke volume
- peak transaortic velocity
- Received October 22, 2015.
- Revision received January 6, 2016.
- Accepted January 12, 2016.
- 2016 American College of Cardiology Foundation
- Nishimura R.A.,
- Otto C.M.,
- Bonow R.O.,
- et al.
- Julius B.K.,
- Spillmann M.,
- Vassalli G.,
- et al.
- Gould K.L.,
- Carabello B.A.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Lanza G.A.,
- Buffon A.,
- Sestito A.,
- et al.
- Bonow R.O.,
- Carabello B.A.,
- Chatterjee K.,
- et al.
- Lansky A.J.,
- Dangas G.,
- Mehran R.,
- et al.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Al-Saadi N.,
- Nagel E.,
- Gross M.,
- et al.
- Park S.J.,
- Enriquez-Sarano M.,
- Chang S.A.,
- et al.
- Rakusan K.,
- Flanagan M.F.,
- Geva T.,
- et al.
- Rajappan K.,
- Rimoldi O.E.,
- Camici P.G.,
- et al.
- Davies J.E.,
- Sen S.,
- Broyd C.,
- et al.
- Costa M.A.,
- Shoemaker S.,
- Futamatsu H.,
- et al.
- Watkins S.,
- McGeoch R.,
- Lyne J.,
- et al.
- John A.S.,
- Dill T.,
- Brandt R.R.,
- et al.
- Kilner P.J.,
- Manzara C.C.,
- Mohiaddin R.H.,
- et al.
- Azevedo C.F.,
- Nigri M.,
- Higuchi M.L.,
- et al.
- Weidemann F.,
- Herrmann S.,
- Stork S.,
- et al.
- Singh A.,
- Ford I.,
- Greenwood J.P.,
- et al.