Author + information
- Received October 11, 2013
- Accepted October 17, 2013
- Published online February 1, 2014.
- Marianna Fontana, MD∗,†,‡,
- Sanjay M. Banypersad, MB ChB∗,†,‡,
- Thomas A. Treibel, MBBS∗,†,
- Viviana Maestrini, MD∗,§,
- Daniel M. Sado, BSc, BM∗,†,
- Steven K. White, BSc, MB ChB∗,†,
- Silvia Pica, MD∗,
- Silvia Castelletti, MD∗,
- Stefan K. Piechnik, PhD, MScEE‖,
- Matthew D. Robson, PhD‖,
- Janet A. Gilbertson, CSci‡,
- Dorota Rowczenio, CSci‡,
- David F. Hutt, BAppSc‡,
- Helen J. Lachmann, MD‡,
- Ashutosh D. Wechalekar, MD‡,
- Carol J. Whelan, MD‡,
- Julian D. Gillmore, MD, PhD‡,
- Philip N. Hawkins, PhD‡ and
- James C. Moon, MD∗,†∗ ()
- ∗The Heart Hospital, London, United Kingdom
- †Institute of Cardiovascular Science, University College London, London, United Kingdom
- ‡National Amyloidosis Centre, University College London, London, United Kingdom
- §Department of Cardiovascular, Respiratory, Nephrologic, Anaesthesiologic and Geriatric Science, La Sapienza University of Rome, Rome, Italy
- ‖Division of Cardiovascular Medicine, Radcliffe Department of Medicine, Oxford Centre for Clinical Magnetic Resonance Research, University of Oxford, John Radcliffe Hospital, Oxford, United Kingdom
- ↵∗Reprint requests and correspondence:
Dr. James C. Moon, The Heart Hospital Imaging Centre, 16-18 Westmoreland Street, London W1G 8PH, United Kingdom.
Objectives The aims of the study were to explore the ability of native myocardial T1 mapping by cardiac magnetic resonance to: 1) detect cardiac involvement in patients with transthyretin amyloidosis (ATTR amyloidosis); 2) track the cardiac amyloid burden; and 3) detect early disease.
Background ATTR amyloidosis is an underdiagnosed cause of heart failure, with no truly quantitative test. In cardiac immunoglobulin light-chain amyloidosis (AL amyloidosis), T1 has high diagnostic accuracy and tracks disease. Here, the diagnostic role of native T1 mapping in the other key type of cardiac amyloid, ATTR amyloidosis, is assessed.
Methods A total of 3 groups were studied: ATTR amyloid patients (n = 85; 70 males, age 73 ± 10 years); healthy individuals with transthyretin mutations in whom standard cardiac investigations were normal (n = 8; 3 males, age 47 ± 6 years); and AL amyloid patients (n = 79; 55 males, age 62 ± 10 years). These were compared with 52 healthy volunteers and 46 patients with hypertrophic cardiomyopathy (HCM). All underwent T1 mapping (shortened modified look-locker inversion recovery); ATTR patients and mutation carriers also underwent cardiac 3,3-diphosphono-1,2-propanodicarboxylicacid (DPD) scintigraphy.
Results T1 was elevated in ATTR patients compared with HCM and normal subjects (1,097 ± 43 ms vs. 1,026 ± 64 ms vs. 967 ± 34 ms, respectively; both p < 0.0001). In established cardiac ATTR amyloidosis, T1 elevation was not as high as in AL amyloidosis (AL 1,130 ± 68 ms; p = 0.01). Diagnostic performance was similar for AL and ATTR amyloid (vs. HCM: AL area under the curve 0.84 [95% confidence interval: 0.76 to 0.92]; ATTR area under the curve 0.85 [95% confidence interval: 0.77 to 0.92]; p < 0.0001). T1 tracked cardiac amyloid burden as determined semiquantitatively by DPD scintigraphy (p < 0.0001). T1 was not elevated in mutation carriers (952 ± 35 ms) but was in isolated DPD grade 1 (n = 9, 1,037 ± 60 ms; p = 0.001).
Conclusions Native myocardial T1 mapping detects cardiac ATTR amyloid with similar diagnostic performance and disease tracking to AL amyloid, but with lower maximal T1 elevation, and appears to be an early disease marker.
Two types of amyloidosis typically infiltrate the ventricular myocardium: immunoglobulin light-chain (AL, or primary systemic) type and transthyretin (ATTR) type. ATTR encompasses senile systemic amyloidosis, in which wild-type transthyretin (TTR) is deposited as amyloid, and hereditary forms, in which genetically variant forms of TTR are implicated. Some TTR variants affect mainly the heart (familial amyloid cardiomyopathy); others mainly the peripheral and autonomic nervous systems in addition to the heart (familial amyloid polyneuropathy). Cardiac ATTR amyloidosis is a progressive and often fatal disorder that may be greatly underdiagnosed and is certainly an underappreciated cause of heart failure in the elderly and specific ethnic populations. ATTR amyloid deposits are present in 8% to 16% of hearts at autopsy in those over age 80 years (1). One particular TTR variant, V122I, which confers susceptibility to amyloid cardiomyopathy, has a population prevalence of 3% to 4% in Afro-Caribbeans (2) and a 10% frequency among individuals of this ethnicity who present with heart failure (3).
Diagnosing cardiac ATTR amyloidosis is often challenging. A suggestive constellation of electrocardiography (ECG), echocardiography, and biomarker findings are found mainly in advanced disease, but interpretation may be confounded by common comorbidities such as left ventricular hypertrophy (LVH), diabetes, diastolic dysfunction, and renal disease (4–6). The challenging diagnosis and lack of validated quantitative investigations to monitor the course of the disease pose unique problems at a time when new specific therapies for ATTR amyloidosis are emerging (7).
New imaging modalities are, however, showing promise. Technetium-labeled bone scintigraphy tracers, notably 3,3-diphosphono-1,2-propanodicarboxylicacid (DPD), localize strikingly to hearts infiltrated by ATTR amyloid (8), whereas cardiac magnetic resonance (CMR) late gadolinium enhancement (LGE) produces a characteristic appearance (9). However, neither is truly quantitative, and both have limitations (10).
Recently, native myocardial T1 mapping has been shown in cardiac AL amyloid to track disease (11). Here, we assess this test in patients with ATTR amyloidosis. We hypothesized that the native myocardial T1 would be elevated in this disease, that elevation would correlate with other disease markers (e.g., intensity of DPD uptake), and that T1 elevation would be an early marker of disease.
Subjects were recruited at the National Amyloidosis Centre, Royal Free Hospital, London, United Kingdom, from 2010 to 2013. A total of 172 individuals were categorized into 3 groups.
ATTR Amyloid Patients
Eighty-five consecutive, consenting patients with cardiac ATTR amyloidosis (70 male; age 73 ± 10 years) were recruited. The presence of cardiac amyloid was defined by presence of ATTR amyloid in a myocardial biopsy or positive DPD scintigraphy. A total of 82% (n = 70) had histological proof of ATTR amyloidosis by Congo red and immunohistochemical staining of myocardial (n = 30, 35%) or other tissues (n = 40, 47%). All patients underwent sequencing of exons 2, 3, and 4 of the TTR gene. No consensus criteria exist for definite cardiac involvement in ATTR amyloid patients, and in this study, definite cardiac ATTR amyloid was defined as: 1) cardiac biopsy showing ATTR amyloid; 2) noncardiac biopsy showing ATTR amyloid in association with LV/right ventricular (RV) thickening in the absence of other explanatory causes; 3) intense DPD uptake in heart (grade 2 or 3 as defined by Perugini et al. ) in the absence of a plasma cell dyscrasia; or 4) noncardiac biopsy showing presence of ATTR amyloid and LGE consistent with cardiac amyloid. In practice, all had apparent LVH. Possible cardiac involvement was defined by minimal cardiac DPD uptake (grade 1 as defined by Perugini et al. ) in the absence of LVH; in practice, none of these patients had LVH.
TTR Gene Carriers
There were 8 TTR gene carriers (n = 8; 3 male, age 47 ± 6 years), defined as individuals with pathogenic TTR gene mutation but no evidence of disease (no cardiac uptake on technetium-99m [99mTc]-DPD scintigraphy and normal echocardiography, CMR, N-terminal pro-brain natriuretic peptide [NT-proBNP], and troponin T). This group constituted the noncardiac involvement group.
AL Amyloid Patients
There were 79 patients with systemic AL amyloid (55 male; age 62 ± 10 years), as proven with biopsies from the myocardium (n = 6, 8%) or other tissues (n = 73, 92%). Cardiac categorization was based on international consensus criteria (12) but with an additional “possible involvement” category. Categorization was defined as: 1) definite cardiac involvement: LV wall thickness of >12 mm in the absence of any other known cause or RV free wall thickening coexisting with LV thickening in the absence of systemic or pulmonary hypertension; 2) possible cardiac involvement: LV wall thickening in the presence of hypertension; RV thickening in the presence of pulmonary hypertension or normal wall thickness with diastolic dysfunction and raised serum biomarkers; and 3) no suspected involvement: normal wall thickness with normal serum biomarkers.
These 3 groups were compared with 46 patients with hypertrophic cardiomyopathy (HCM) and 52 healthy volunteers.
There were 46 patients with HCM (n = 46; age 50 ± 13 years, 34 male) fulfilling diagnostic criteria (13). A total of 72% of patients had an asymmetrical septal hypertrophy pattern (the remainder had apical predominant hypertrophy); 60% had LV outflow tract obstruction; and 76% were found to have LGE in a variety of locations, such as at the RV insertion points or the LV apex.
A total of 52 healthy volunteers (n = 52; age 46 ± 15 years, 17 male) were recruited through advertising in hospitals, universities, and general practitioner surgeries. All had no history or symptoms of cardiovascular disease or diabetes mellitus, and all had normal 12-lead ECG and normal CMR scan. No patient was on cardioactive medication, except for 4 patients on statins for primary prevention.
All patients and healthy controls underwent 12-lead ECG. Cardiac amyloid patients/carriers additionally underwent assays of cardiac biomarkers (NT-proBNP and troponin T), echocardiography, and a 6-min walk test when health and patient choice permitted (e.g., test was not undertaken when prohibited by arthritis, postural hypotension, or neuropathy). The ATTR group also underwent DPD scintigraphy. The baseline characteristics of all patients are provided in Table 1. All ethics were approved by the UCL/UCLH Joint Committees on the Ethics of Human Research Committee.
All patients with contraindications to CMR (i.e., glomerular filtration rate <30 ml/min, incompatible devices) were excluded from the study.
All subjects underwent standard CMR on a 1.5-T clinical scanner (Avanto, Siemens Healthcare, Erlangen, Germany). A standard volume and LGE study was performed (14). The contrast agent was 0.1 mmol/kg of gadolinium-based contrast (gadoterate meglumine [Dotarem, Guerbet SA, Paris, France]) with 5-min delay, and the LGE sequence was either a standard fast low-angle single shot inversion recovery or true fast imaging with steady state free precession sequence, with a phase sensitive inversion recovery sequence or magnitude reconstruction. For native T1 mapping, basal and midventricular short-axis and 4-chamber long-axis views were acquired using the shortened modified look-locker inversion recovery (ShMOLLI) sequence after regional shimming, as previously described (Fig. 1) (15).
Patients were scanned using 2 GE Medical Systems (Fairfield, Connecticut) hybrid single photon emission computed tomography (SPECT) computed tomography (CT) gamma cameras (Infinia Hawkeye 4 and Discovery 670) following administration of 700 MBq of intravenously-injected 99mTc-DPD. The 3-h (delayed) whole body planar images were acquired, followed by SPECT of the heart with a low-dose, noncontrast CT scan. Gated/nongated cardiac SPECT reconstruction and SPECT-CT image fusion was performed on the GE Xeleris workstation. Cardiac retention of 99mTc-DPD was visually scored as:
• Grade 0: no visible myocardial uptake in both the delayed planar or cardiac SPECT-CT scan;
• Grade 1: cardiac uptake on SPECT-CT only or cardiac uptake of less intensity than the accompanying normal bone distribution;
• Grade 2: moderate cardiac uptake with some attenuation of bone signal; and
• Grade 3: strong cardiac uptake with little or no bone uptake.
CMR image analysis
All CMR images and maps were analyzed offline. For T1 measurements, the basal ventricular short-axis or the 4-chamber ShMOLLI image was manually contoured approximately 2 pixels in (to minimize partial volume effects) from the endocardium and epicardium, and the average T1 value was calculated (Fig. 1). This was drawn without review of the LGE images. The LGE images were visually analyzed for the presence or absence of enhancement, blinded to T1 mapping results. The presence of LGE was classified as: circumferential in the subendocardium; diffuse circumferential (extend into the epicardial layer); and abnormal contrast handling on T1 scout with no discernible LGE.
Statistical analysis was performed using IBM SPSS Statistics version 19 (IBM, Somers, New York). All continuous variables were normally distributed (Shapiro-Wilk) other than NT-proBNP and troponin T, which were therefore log-transformed for bivariate testing; these are presented as mean ± SD, with nontransformed NT-proBNP presented as median and interquartile range. Comparisons between groups were performed by 1-way analysis of variance with post-hoc Bonferroni correction. The chi-square test or Fisher exact test was used to compare discrete data as appropriate. Receiver-operating characteristic (ROC) curve analysis was performed to define the diagnostic accuracy of native T1. Correlation between continuous variables was assessed using Pearson's r correlation coefficient. Statistical significance was defined as p < 0.05.
A total of 85 patients with ATTR amyloid, 8 TTR mutation carriers and 79 patients with AL amyloid were enrolled. These were compared with 52 healthy volunteers and 46 patients with HCM. Baseline characteristics are shown in Table 1. Amyloid patients had the following comorbidities: treated hypertension (22% ATTR, 15% AL); diabetes (12% ATTR, 3% AL); and angiographically-confirmed coronary artery disease (13% ATTR, 9% AL). The echocardiogram was performed within 1 day of the CMR. The time gap between the DPD and CMR was 26 ± 36 days. Thirty-five ATTR amyloid patients were familial (V122I [n = 18], T60A [n = 6], V30M [n = 2], and E54G [n = 2], and all others unique: D38Y, G47V, E89K, I84S, I107F, L12P, and S77Y); 50 had senile systemic amyloidosis (SSA). Of the 8 gene carriers, 5 had TTR V30M and 3 had T60A. Compared with definite cardiac AL patients, definite ATTR amyloid patients had a higher LV mass index (133 ± 27 vs. 101 ± 25) and reduced ejection fraction (53 ± 15% vs. 61 ± 11%). The PR interval and QRS were longer in ATTR (PR 209 ± 54 ms vs. 185 ± 38 ms, QRS 116 ± 28 ms vs. 106 ± 23 ms; both p < 0.05). The NT-proBNP and troponin T biomarker concentrations were similar.
T1 was elevated in ATTR patients compared with HCM and normal subjects (1,097 ± 43 ms vs. 1,026 ± 64 ms vs. 967 ± 34 ms; both p < 0.0001). In established disease, ATTR T1 elevation was not as high as in AL (AL 1,130 ± 68 ms; p = 0.01). T1 was not elevated in mutation carriers (952 ± 35 ms) (Figs. 2 and 3).
T1 diagnostic accuracy
The ROC curve analysis was performed for the discrimination of possible or definite cardiac amyloid from the meaningful combined differentials of HCM, systemic amyloid without detected cardiac involvement, or ATTR mutation-positive patients without evidence of cardiac amyloid. Using ROC analysis, ATTR and AL amyloid patients with possible or definite cardiac involvement had an area under the receiver-operating characteristic curve (AUC) of 0.85 (95% confidence interval [CI]: 0.79 to 0.92).
Example cut-off values to diagnose cardiac amyloid (high specificity) are 1,048 ms, 1,065 ms, and 1,090 ms. These values have 80%, 85%, and 90% specificity and 83%, 74%, and 56% sensitivity, respectively. Example cut-off values to rule out cardiac amyloid (high sensitivity) are 954 ms, 968 ms, and 1,012 ms. These values have 99%, 98%, and 95% sensitivity and 17%, 30%, and 58% specificity, respectively. Native T1 was similarly accurate for AL and ATTR (AUC: 0.84, 95% CI: 0.76 to 0.91 and AUC 0.85, 95% CI: 0.77 to 0.92, respectively) (Fig. 4).
T1 and DPD/LGE findings
T1 increased with increasing cardiac amyloid burden, as assessed by bone scintigraphy (p < 0.0001 for trend) (Fig. 5). T1 was not elevated in mutation carriers (952 ± 35 ms) but was elevated in the 9 patients with isolated DPD grade 1 (1,037 ± 60 ms, p = 0.001), all of which had no amyloid-like LGE (but 1 had inferior myocardial infarction and 1 had RV LGE).
T1 and cardiac function, biomarkers, ECG, and 6-min walk test
Correlations were broadly similar for AL and ATTR disease (Table 2). T1 correlated with indexes of systolic and diastolic function, indexed LV mass, and known prognostic biomarkers both in ATTR and AL amyloid patients. In ATTR patients, T1 correlated with indexed left atrial area, 6-min walk test performance, and PR and QRS duration on ECG, whereas in AL patients, T1 correlated with indexed stroke volume, ECG, limb lead mean voltage, and E deceleration time (Table 2).
In this, the largest ever CMR study in patients with amyloidosis, we found that native myocardial T1 mapping has a high diagnostic accuracy for cardiac amyloid for both AL and ATTR when compared against HCM, a relevant clinical differential diagnosis. Furthermore, T1 tracks cardiac amyloid burden in both diseases, and is more sensitive for detecting early disease in gene mutation carriers than LGE imaging. In both amyloid types, T1 tracks markers of systolic and diastolic function, mass, and prognostic markers. In ATTR amyloid, T1 additionally correlates with ECG PR and QRS duration and indexed left atrial area, whereas in AL type, it correlates with reductions in limb lead voltages. T1 also has functional associations with a reduction in 6-min walk test in ATTR amyloidosis. Interestingly and perhaps unexpectedly (16), T1 elevation was lower in ATTR compared with AL type.
Amyloidosis is considered the exemplar of an interstitial disease, as the quantity of amyloid in the extracellular space amounts to kilograms overall in some patients and is able to constitute the majority of the heart by weight at times (17).
Our earlier work in AL amyloidosis demonstrated that measurement of myocardial T1 times using ShMOLLI had high diagnostic accuracy (against aortic stenosis) and tracks disease burden (11). Here, this work is extended to ATTR, and the diagnostic accuracy was tested against hypertrophic cardiomyopathy, a relevant clinical differential. Although T1 was raised in the ATTR amyloid, it was not as high as in AL type, a surprising finding given that ventricular wall thickness is greater in ATTR amyloid (16). T1 mapping measures a composite tissue signal from both cells and the interstitium. The extent and/or distribution of amyloid, plus how it interacts with water or changes in the myocyte signal, could all be implicated in causing the T1 difference. Therefore, the less raised T1 value found in patients with ATTR amyloidosis could be due to (amongst others), a lower amyloid burden, less hydration of the amyloid, less collagen associated with amyloid, or differential effects on the intracellular signal. Finally, AL may have additional processes occurring, such as edema from possible light chain toxicity (18,19). Further work is needed. Many differences in the biology of ATTR and AL cardiac amyloidosis have been described but are not understood. This study shows specific differences between AL and ATTR in their correlations with other parameters; for example, the positive correlations in ATTR amyloid of T1 with left atrial area, PR, and QRS duration, and the negative correlations in AL amyloid of T1 with mean QRS voltage in the limb leads. These findings support the concept that ATTR amyloid may be a more purely infiltrative disease, whereas AL may have a dual pathology with contributions from interstitial expansion and cell death.
Native myocardial T1 yielded high diagnostic accuracy in ATTR and AL amyloidosis against the common clinical differential diagnosis of HCM. These findings combine a number of key advantages of the mapping technique: the absence of need for contrast, a single breath-hold per T1 map, simple analysis, and the potential for measurement of whole-heart T1. Other investigators have proposed bone scintigraphy using the DPD tracer as the noninvasive “gold standard” for diagnosis of cardiac ATTR amyloid (20). However, this is semiquantitative and is scored in 4 grades based on visual estimation; although the radiation dose is low, serial follow-up is problematic for gene mutation carriers. Native T1 showed high concordance with DPD scintigraphy, and T1 was measured on a continuous scale, which is a possible advantage. DPD scanning dichotomizes mutation carriers into DPD negative or DPD grade 1. T1 was exactly concordant with this dichotomy, with T1 elevation and DPD grade 1 patients having no other abnormalities (no hypertrophy, no biomarker elevation, and no amyloid-related LGE), suggesting that both methods identify the earliest disease expression. These preliminary findings require further confirmation in larger, particularly multicenter, studies with assessment of diagnostic and prognostic performance. In native T1 measurement, there is the potential for age and sex biases. Some of these biases may be about partial voluming and may have reduced relevance in hypertrophied hearts. The disease biology means that the ATTR and AL populations have a different age of presentation, with ATTR patients being older and more often male. The HCM patients and the controls are younger. Data on age-dependent changes in T1 are conflicting. A subtle trend in increasing T1 measurement has been detected in MESA (Multi-Ethnic Study of Atherosclerosis) (r2 = 0.021, 5 ms/decade in men but not women ), but there was a reduction in T1 with age in women but not men with ShMOLLI (22). Fortunately, most of the effect sizes detected here are large compared to the potential biases, and some of them are in the opposite direction of a confounding bias: for example, the (10 years younger) AL patients have more T1 elevation than the ATTR patients by 33 ms, which is the opposite of the age-related bias that most experts expect to exist. The ATTR patients were 82% male, whereas the AL population was 72% male. A male/female difference in T1 was not found in patients over age 45 years in the paper by Piechnik et al. (22), but even if a 24 ms difference was found in the young males versus females, this would make only a ∼3 ms overall difference, ∼10% of the difference found in this paper. Furthermore, ATTR patients more often had hypertension, coronary artery disease, and diabetes. All of these factors might be expected to contribute to an increase in T1 times in ATTR patients—the opposite trend to that found, which could potentially lead to an underestimation of the difference between AL and ATTR patients.
The main limitation of this study is the lack of histological validation of the technique. This is challenging because diagnosis in many patients was secured through biopsy of noncardiac sites, and it was not ethical to obtain cardiac biopsies for research given its invasive nature. Furthermore, quantification of amyloid (rather than presence/absence) is challenging in myocardial biopsies due to its often patchy and microscopic distribution. The second limitation is lack of biopsy evidence of amyloid in 18% of ATTR patients. However, this cohort of patients was fully characterized with all other clinical investigative techniques currently available. When combined, these are known to provide high diagnostic accuracy. CMR extracellular volume measurement was not included in this study. No follow-up data are available at this time. T1 mapping is yet to be standardized across manufacturers, although efforts are underway to do this. In our paper, we analyzed only a single region of interest in the septum. A segmental or whole-heart approach should be explored in the future.
Native myocardial T1 mapping detects cardiac ATTR amyloid and has similar performance for diagnosis and tracking disease in both ATTR and AL amyloidosis. The lower T1 elevation in ATTR amyloid, and the specific differences between AL and ATTR correlations with other cardiac findings, may support a concept of ATTR amyloid as a more purely infiltrative disease, whereas the AL type may have a dual pathology with both interstitial expansion and cell death.
The authors are thankful for the contributions of the administrative and nursing staff, histopathologists, geneticists, echocardiographers, and radiographers at the National Amyloidosis Centre and the Heart Hospital.
This paper was partially funded by the British Heart Foundation. Dr. Piechnik's institution has a research agreement with Siemens Medical. Drs. Piechnik and Robson have U.S. patents pending for cardiac gated mapping of T1 (U.S. patent pending 61/387,591; all rights sold exclusively to Siemens Medical) and a color map design method for cardiovascular T1 mapping images (U.S. patent pending 61/689,067). Dr. Moon has received grant funding from GlaxoSmithKline. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- AL amyloidosis
- immunoglobulin light-chain amyloidosis
- ATTR amyloidosis
- transthyretin amyloidosis
- cardiac magnetic resonance
- hypertrophic cardiomyopathy
- late gadolinium enhancement
- left ventricle/ventricular
- left ventricular hypertrophy
- N-terminal pro-brain natriuretic peptide
- right ventricle/ventricular
- shortened modified look-locker inversion recovery
- transthyretin protein
- Received October 11, 2013.
- Accepted October 17, 2013.
- American College of Cardiology Foundation
- Buxbaum J.,
- Jacobson D.R.,
- Tagoe C.,
- et al.
- Pinney J.H.,
- Whelan C.J.,
- Petrie A.,
- et al.
- Ruberg F.L.,
- Berk J.L.
- Perugini E.,
- Guidalotti P.L.,
- Salvi F.,
- et al.
- Maceira A.M.,
- Joshi J.,
- Prasad S.K.,
- et al.
- Karamitsos T.D.,
- Piechnik S.K.,
- Banypersad S.M.,
- et al.
- Gertz M.A.,
- Comenzo R.,
- Falk R.H.,
- et al.
- Elliott P.,
- Andersson B.,
- Arbustini E.,
- et al.
- Ruberg F.L.
- Banypersad S.M.,
- Sado D.M.,
- Flett A.S.,
- et al.
- Dubrey S.W.,
- Cha K.,
- Skinner M.,
- LaValley M.,
- Falk R.H.
- Shi J.,
- Guan J.,
- Jiang B.,
- et al.
- Liu C.Y.,
- Liu Y.C.,
- Wu C.,
- et al.