Author + information
- Received April 7, 2016
- Accepted May 10, 2016
- Published online August 16, 2016.
- Sunil V. Rao, MDa,∗ (, )
- Uwe Zeymer, MDb,
- Pamela S. Douglas, MDa,
- Hussein Al-Khalidi, PhDa,
- Jennifer A. White, MSa,
- Jingyu Liu, PhDc,
- Howard Levy, MD, PhDc,
- Victor Guetta, MDd,
- C. Michael Gibson, MDe,
- Jean-Francois Tanguay, MDf,
- Paul Vermeersch, MDg,
- Jérôme Roncalli, MD, PhDh,
- Jaroslaw D. Kasprzak, MDi,
- Timothy D. Henry, MDj,
- Norbert Frey, MDk,
- Oscar Kracoff, MDl,
- Jay H. Traverse, MDm,
- Derek P. Chew, MBBSn,
- Jose Lopez-Sendon, MDo,
- Reinilde Heyrman, MDc and
- Mitchell W. Krucoff, MDa
- aDuke Clinical Research Institute, Durham, North Carolina
- bHerzzentrum Luwidshafen, Ludwigshafen, Germany
- cBellerophon Therapeutics Inc., Hampton, New Jersey
- dSheba Tel Ha Shomer Hospital, Tel Ha Shomer, Israel
- eBeth Israel Deaconess Medical Center, Boston, Massachusetts
- fMontreal Heart Institute, Montreal, Quebec, Canada
- gFree University, Brussels, Belgium
- hRangueil University Hospital, Toulouse, France
- iMedical University of Lodz, Lodz, Poland
- jCedars-Sinai Heart Institute, Los Angeles, California
- kUniversitätsklinikum Schleswig-Holstein, Campus Kiel Klinik für Kardiologie und Angiologie, Kiel, Germany
- lKaplan Medical Center, Rehovot, Israel
- mMinneapolis Heart Institute, Minneapolis, Minnesota
- nFlinders University, Adelaide, South Australia, Australia
- oHospital Universitario La Paz, Madrid, Spain
- ↵∗Reprint requests and correspondence:
Dr. Sunil V. Rao, Department of Cardiology, Duke Clinical Research Institute, 508 Fulton Street (111A), Durham, North Carolina 27705.
Background Bioabsorbable cardiac matrix (BCM) is a novel device that attenuates adverse left ventricular (LV) remodeling after large myocardial infarctions in experimental models.
Objectives This study aimed to analyze whether BCM, compared with saline control, would result in less LV dilation and fewer adverse clinical events between baseline and 6 months.
Methods In an international, randomized, double-blind, controlled trial, 303 subjects with large areas of infarction despite successful primary percutaneous coronary intervention (PCI) of ST-segment elevation myocardial infarction (STEMI) were randomized 2:1 to BCM or saline injected into the infarct-related artery 2 to 5 days after primary PCI. The primary outcome was mean change from baseline in LV end-diastolic volume index (LVEDVI) at 6 months. Secondary outcomes included change in Kansas City Cardiomyopathy Questionnaire score, 6-minute walk time, and New York Heart Association functional class at 6 months. The primary safety endpoint was a composite of cardiovascular death, recurrent MI, target-vessel revascularization, stent thrombosis, significant arrhythmia requiring therapy, or myocardial rupture through 6 months.
Results In total, 201 subjects were assigned to BCM and 102 to saline control. There was no significant difference in change in LVEDVI from baseline to 6 months between the groups (mean change ± SD: BCM 14.1 ± 28.9 ml/m2 vs. saline 11.7 ± 26.9 ml/m2; p = 0.49). There was also no significant difference in the secondary endpoints. The rates of the primary safety outcome were similar between the 2 groups (BCM 11.6% vs. saline 9.1%; p = 0.37).
Conclusions Intracoronary deployment of BCM 2 to 5 days after successful reperfusion in subjects with large myocardial infarction did not reduce adverse LV remodeling or cardiac clinical events at 6 months. (IK-5001 for the Prevention of Remodeling of the Ventricle and Congestive Heart Failure After Acute Myocardial Infarction [PRESERVATION I]; NCT01226563)
Despite reduced mortality via timely intervention and antithrombotic therapies, some patients still experience large ST-segment elevation myocardial infarctions (STEMI) related to delayed presentation, unsuccessful reperfusion, microvascular obstruction, or reperfusion injury. Such patients subsequently experience pathological left ventricular (LV) remodeling associated with functional impairment and chronic heart failure (HF). This is also the result of extracellular matrix (ECM) degradation with loss of tissue integrity, which leads to thinning of the infarct zone. In this setting, specific anatomic features such as higher left ventricular end-diastolic volume index (LVEDVI) are associated with adverse prognosis (1,2). Thus, LV pathological remodeling contributes independently to HF progression. Once HF manifests, the mortality rate is 25% to 30% within 1 year and 50% within 5 years (3). Therefore, preventing LV remodeling after STEMI has the potential to reduce HF and improve survival.
Two events occur in the setting of large myocardial infarctions (MIs): apoptosis and ECM degradation with high concentrations of extracellular ionized calcium. One novel strategy to prevent pathological remodeling leverages an injectable bioabsorbable alginate, or bioabsorbable cardiac matrix (BCM), which provides temporary structural support to the infarct zone until mature fibrotic tissue develops (4). Alginate is derived from seaweed and has been widely used in medical applications and foodstuffs. A small study of direct intramyocardial injection of the alginate polymer Algisyl (LoneStar Heart, Inc., Dallas, Texas) via thoracotomy in patients with advanced HF demonstrated improved exercise capacity, symptoms, and clinical outcomes at 12 months (5). Because BCM can extravasate through permeable capillaries to reach the extracellular space, it can be administered intravascularly. In the presence of ionized calcium present in apoptotic tissues, BCM cross-links into a flexible biological ECM-like scaffolding that creates a cast for the tissue and is gradually and completely degraded and excreted through the kidneys in 3 to 6 months. In preclinical STEMI models, BCM replaced damaged ECM; structurally supported damaged tissue; prevented local dyskinesis; and increased LV thickness, which led to reduced wall stress, improved LV function, and prevention of progressive pathological remodeling (6).
We conducted the PRESERVATION I (Prevention of Remodeling of the Ventricle and Congestive Heart Failure After Acute Myocardial Infarction) trial to determine the safety and effectiveness of intracoronary deployment of BCM on measures of LV remodeling, clinical outcomes, and functional status in patients with large infarctions several days after successful primary or rescue percutaneous coronary intervention (PCI) for STEMI. We hypothesized that BCM, compared with saline control, would result in less LV dilation and fewer adverse clinical events between baseline and 6 months.
PRESERVATION I was a multicenter, randomized, double-blind, placebo-controlled trial. Design details have been published (7). National and institutional regulatory authorities and ethics committees approved the trial design, and all subjects provided written informed consent. Trial committee members and investigators are listed in the Online Appendix.
BCM is an aqueous mixture of 1% sodium alginate and 0.3% calcium gluconate. It is a sterile, colorless liquid that is not cytotoxic or mutagenic. Its mechanism of action involves assembling into a flexible gel that structurally resembles ECM when exposed to excess ionized calcium present in infarcted myocardium (8). BCM is designated as a medical device as defined by the U.S. Food and Drug Administration and other participating regulatory authorities.
Subjects were eligible for enrollment in PRESERVATION I if they were ≥18 years old, had undergone successful primary PCI for STEMI, and had a large MI and ≥1 of the following: delayed PCI presentation >6 h from symptom onset, significant new Q waves in ≥2 anterior leads or anterior ST-segment elevation of ≥3 mm persistent 24 h after PCI, or new onset of congestive HF or cardiogenic shock persistent at 24 h after PCI, and documentation of a large infarction by imaging. Subjects undergoing rescue PCI after fibrinolysis were eligible as long as the PCI was successful. Detailed inclusion and exclusion criteria are presented in the Online Appendix.
Two to 5 days after the index STEMI was successfully treated with primary or rescue PCI, eligible subjects underwent cardiac catheterization to determine TIMI (Thrombolysis In Myocardial Infarction) flow in the infarct-related artery. Subjects with confirmed TIMI flow grade 3 in the infarct-related artery were randomized in a 2:1 ratio to BCM or saline by use of a computerized central interactive voice and Internet response system. Randomization was stratified by baseline ejection fraction (<35% or ≥35%).
A total of 4 ml of BCM or saline control was subselectively injected into the infarct artery through a microcatheter in 30 to 60 s, followed by a saline flush. The 4 ml dose of BCM was determined empirically after the first-in-humans study. A final angiogram to confirm the integrity of the artery, stent, and flow was taken after all equipment except the guide catheter was removed. Guidelines-based treatment for STEMI was recommended for all subjects.
Effectiveness and safety endpoints
The primary effectiveness endpoint was the change in LVEDVI, measured by transthoracic echocardiography from baseline to 6 months after BCM deployment. Resting transthoracic echocardiograms were performed at baseline; 0 to 18 h post-deployment; and at 1, 3, 6, and 12 months after enrollment. This paper reports the primary 6-month endpoint. An independent, blinded echocardiographic core laboratory (Duke Clinical Research Institute, Durham, North Carolina) oversaw image acquisition and performed all measurements according to guideline recommendations and best practices (9). All sites were qualified by the core laboratory before enrollment, and only trained and certified echocardiographers performed the baseline and 6-month assessments. All studies were measured by an experienced research sonographer and overread by a level III trained cardiologist. Reproducibility testing for interobserver variability showed intraclass correlation coefficients of 0.92 for 3-dimensional (3D) end-diastolic volume and 1.00 for 2-dimensional end-diastolic volume.
Secondary effectiveness endpoints were subject-reported functional and quality-of-life assessments using the Kansas City Cardiomyopathy Questionnaire (KCCQ), 6-min walk test (6MWT), and New York Heart Association (NYHA) functional classification. We also assessed clinical outcomes.
The primary safety endpoint was a composite of cardiovascular death, recurrent MI, target-vessel revascularization, stent thrombosis (defined according to the Academic Research Consortium), significant arrhythmia requiring therapy, or myocardial rupture through 6 months. We also assessed the safety of the deployment procedure using 24-h periprocedural digital 12-lead Holter monitoring to determine ischemic ST-segment changes, life-threatening arrhythmias, and QTc duration. Holter recordings were analyzed in an independent core laboratory (Duke Clinical Research Institute) blinded to treatment assignment. An independent, blinded angiographic core laboratory (PERFUSE, Boston, Massachusetts) analyzed all procedural angiograms for TIMI flow and perfusion. An independent clinical events committee used standardized definitions to adjudicate clinical outcomes; these definitions are provided in the Online Appendix.
The sample size was based on an estimated change from baseline to 6 months in the LVEDVI after device deployment. With 276 subjects with paired baseline and 6-month echocardiograms, and with a device allocation ratio of 2:1 (BCM to saline control), the trial had 80% power to detect a treatment difference of ≥5 ml/m2 in LVEDVI between BCM and saline control under a significance level of 0.05. On the basis of human pilot data with BCM, the common SD was estimated at 13.89 ml/m2. Additional subjects were recruited to account for potential dropouts. All analyses were conducted on the basis of the intention-to-treat principle.
For the primary effectiveness endpoint, if paired baseline and 6-month 3D echocardiography was not available, paired biplane 2-dimensional echocardiography was used. Subjects who died or required an LV assist device, cardiac resynchronization therapy, or automated implantable cardioverter-defibrillator placement or heart transplantation during the study had their worst LVEDVI value (i.e., the worst score of all nonmissing assessments) assigned for the follow-up effectiveness assessment. If a subject missed a primary effectiveness assessment at a follow-up visit, the last available assessment was used to impute the missing value for the primary endpoint. Missing data imputations for the secondary effectiveness endpoints (KCCQ, 6MWT, and NYHA functional class) followed the same approach. In addition, for subjects who could not perform the 6MWT at baseline, baseline values were imputed by the lower quartile of the baseline values from those who did perform the 6MWT. Two additional analyses for 6MWT were performed: 1) Subjects who could not perform the 6MWT were assigned 0 meters; and 2) a test was performed among subjects who could walk at baseline and at 6 months.
All statistical tests were 2-sided with a significance level of 0.05. To control for type 1 error at a significance level of 0.05, the statistical tests for the secondary effectiveness endpoints were to be performed only if the statistical test for the primary effectiveness endpoints was significant. Furthermore, a hierarchical gate-keeping procedure was applied to test the treatment difference for each secondary effectiveness endpoint in order of: 1) KCCQ; 2) 6MWT; 3) NYHA functional classification; 4) cardiovascular death or nonfatal HF events or cardiovascular hospitalizations; and 5) time to first rehospitalization because of any cardiovascular event. Each test was evaluated at the 0.05 level. The procedure was stopped if lack of significance (p > 0.05) was found, and the remaining tests in the sequence were to be considered nonsignificant.
Finally, the effectiveness endpoint of LVEDVI was analyzed with a mixed-effects model with repeated measures at 1-, 3-, and 6-month follow-up visits. Fixed factors of the model included the randomization strata, device indicator, and device-indicator-by-visit interaction. The follow-up visit was treated as a random repeated factor and the baseline value as a covariate in the model.
We randomized 303 subjects from 61 sites in 9 countries from April 2012 to December 2014 (Figure 1). Follow-up ended in June 2015 when data on the 6-month primary endpoint were available for all but 5 subjects. There were several baseline differences between subjects assigned to BCM versus saline control: BCM subjects were younger, had higher body weight, more often had a history of hypertension and diabetes mellitus, and had higher baseline N-terminal pro-B-type natriuretic peptide (Table 1). Subjects were randomized at a mean of 3.4 days after primary PCI in both groups (SD ± 0.97 for BCM and ± 0.91 for saline). Measures of infarct size are shown in Online Table 1.
The infarct-related artery was the left anterior descending artery in the vast majority of cases. More than 50% of subjects underwent transradial primary PCI, and 65% received a drug-eluting stent (Table 1).
Most subjects had the protocol-specified 4 ml of study device deployed in the infarct-related artery (Table 1); 6 subjects (n = 5 BCM; n = 1 saline) had <4 ml deployed, and 11 subjects (n = 7 BCM; n = 4 saline) had >4 ml deployed. There were no significant differences between groups with respect to pre-procedure or post-procedure TIMI flow grade, frame count, or myocardial perfusion grade (Online Table 2).
Effectiveness and safety endpoints
There was no significant difference in LVEDVI at 6 months between BCM and saline control (Table 2); the mean difference from baseline to 6 months was a 2.4 ml/m2 (95% confidence interval [CI]: −4.4 to 9.2 ml/m2) increase with BCM. Neither were there any significant differences in secondary effectiveness endpoints: For BCM, there was a trend toward a benefit in change in 6MWT from baseline to 6 months, but it did not reach statistical significance (p = 0.051) for the primary analysis (Table 2) or when a value of 0 meters was imputed for subjects who could not walk. There were 129 patients assigned to BCM and 67 assigned to saline control who could walk at baseline and at 6 months. When the analysis of 6MWT was limited to these patients, there was a significant improvement in the BCM group (mean change in 6MWT: BCM 149.4 m vs. saline 94.6 m; p = 0003). Results of the mixed-effects model showed there was a 3.34 ml/m2 (95% CI: −0.79 to 7.46 ml/m2) greater increase in LVEDVI from baseline to 6 months in the BCM group than in the saline group (p = 0.11).
There was no significant difference between BCM and saline control in the primary safety evaluation from baseline to 6 months (Central Illustration). Regarding individual safety endpoints, there were numerically lower rates of death and MI at 6 months with BCM; however, there were numerically higher rates of stent thrombosis, target-vessel revascularization, and significant arrhythmia requiring therapy with BCM (Table 3). Continuous Holter monitoring revealed no difference in the incidence of periprocedural ischemic ST-segment changes (BCM 9.0%; saline 7.8%) or arrhythmias between BCM and saline control. There was also no difference in the median QTc interval post-deployment (BCM 413.8 ms; saline 417.6 ms).
The incidence of any adverse event was similar between groups (BCM 76.1%; 74.5% saline). The rate of serious adverse events was 46.8% with BCM and 37.3% with saline. No subject in either group had a serious adverse event that was considered definitely related to the study device by the investigator. Serious adverse events that were considered possibly related to the study device occurred numerically more frequently with BCM (5.0%) than with saline (2.9%). The proportion of subjects with a serious adverse event that was considered definitely related to the required deployment procedure was slightly higher with BCM (1.5%) than with saline (0%), as was the proportion of serious adverse events considered possibly related to deployment procedures (4.0% vs. 2.0%).
In this trial, intracoronary injection of BCM 2 to 5 days after large STEMI did not attenuate increases in LVEDVI or major adverse cardiac clinical events through 6 months compared with saline control. Several medical therapies directed at reducing infarct size by attenuating reperfusion injury have failed to improve either anatomic measures or clinical outcomes (10,11). BCM was not designed to reduce infarct size; its mechanism of action relies on large areas of infarction and attendant high levels of extracellular ionized calcium that allow it to form a gel that resembles ECM. Through gelation, BCM was designed to tether infarcted myocardium and reduce ventricular dilation. Pre-clinical studies in rats (4) indicated that intramyocardial injection of BCM within 7 days of MI and intracoronary injection in dogs and pigs attenuated both LV dilation and decrease in LV function (data on file, Bellerophon Therapeutics Inc., Warren, New Jersey). In a nonrandomized, first-in-human study of 27 subjects with moderate-to-large STEMI who had 2 ml of BCM injected into the infarct-related artery within 7 days after MI (12), there were few adverse events reported, all of which an independent data safety monitoring board deemed consistent with expected sequelae of large infarctions.
PRESERVATION I was designed to follow this pilot study and test whether the mechanism of action of BCM would reduce adverse LV remodeling and improve clinical outcomes. The trial included subjects with large infarctions and was conducted in a double-blind fashion, including a sham procedure using saline control, with LVEDVI as the primary endpoint. Given that LV end-systolic volume index and LVEDVI are both strongly related to outcomes (2,13,14), we chose to use LVEDVI because BCM is designed to provide a myocardial scaffold to reduce post-infarction dilation. This is more directly measured with diastolic rather than systolic volume, which incorporates both size and systolic shortening. In addition, by enrollment criteria, all patients had very recent, very large MIs and had been revascularized before study enrollment. The extent to which the myocardium at risk was still stunned at the time of the baseline echocardiogram was unknown; however, such stunning would be much more likely to affect systolic function and LV end-systolic volume index than LVEDVI. The use of LVEDVI as the endpoint would remove much if not all of the potential confounding influence of recovery from ischemia on the trial’s endpoint.
Another important feature of PRESERVATION I was the use of 3D echocardiography to assess the primary endpoint, which allowed accurate assessment of LVEDVI with an imaging modality that is widely available. Finally, safety was rigorously assessed with blinded independent laboratories, continuous monitoring, and serial angiograms before and after deployment for evidence of change in TIMI flow. BCM was associated with a numerical increase in stent thrombosis compared with saline control. Each stent thrombosis event was reviewed by the data safety monitoring board, which recommended continuation of the trial; however, although the difference in stent thrombosis did not reach statistical significance, future studies of BCM will require close scrutiny to ensure that deployment of BCM does not significantly increase intracoronary thrombotic events.
Several other aspects of the study deserve comment. There were some baseline differences between the 2 randomized groups; however, none of these reached statistical significance, which indicates good balance between patients assigned to BCM and those assigned to saline control. In preclinical studies, BCM replaced areas of ECM, provided flexible scaffolding for the infarcted myocardium, and improved anatomic measures of LV remodeling; however, in PRESERVATION I, BCM failed to provide any measureable anatomic or clinical benefit in subjects with large infarctions.
This study has several limiting factors, including a lack of benefit. First, the volume of BCM might not have been high enough for the large infarctions included in this trial. The 4 ml deployment volume was based on the first-in-human study demonstrating tolerability and safety of 2 ml of BCM. Whether a larger volume could show benefit would require future study. Second, our study population might have been a limiting factor: In subjects with very large infarctions, it is possible that remodeling simply cannot be prevented. Among subjects who had infarct size measured, approximately 30% of the LV was involved. Given the large size of the infarcted myocardium, it is also possible that attendant microvascular occlusion prevented BCM from reaching the intracellular space in the infarct zone. This might also explain the difference between PRESERVATION I and the AUGMENT-HF (Algisyl-LVR as a Method of Left Ventricular Augmentation for Heart Failure) trial, in which the alginate polymer Algisyl was directly injected into the myocardium (5). However, PRESERVATION I, like AUGMENT-HF, did show a benefit of alginate on functional outcomes such as exercise capacity. BCM might be better suited for those with slightly smaller infarcts who are still at risk for adverse LV remodeling, a decidedly narrower group.
Third, the trial did not test whether more acute deployment of BCM (e.g., during primary PCI) might vary its impact. Although it is possible that BCM could have a benefit with earlier administration, many people with STEMI with large areas at risk do not actually go on to have large infarctions, which makes the study of the mechanistic benefit of BCM more complex in such a mixed cohort. In PRESERVATION I, we deployed BCM in a procedure separate from the primary PCI, which allowed for comprehensive assessment of infarct size. This was designed to include a population with a greater likelihood of benefit from a therapy aimed at preventing LV remodeling. In fact, an unanticipated consequence of this design included a modest number of major adverse events from the repeat catheterization, even though this was unrelated to BCM versus saline deployment per se.
In the PRESERVATION I trial, testing the effect of a novel intracoronary alginate in subjects with large infarctions despite successful reperfusion of STEMI, BCM deployment did not reduce LV remodeling or major cardiovascular events through 6 months. However, its effects on functional outcomes such as the 6MWT suggest that further investigation of alginate in the setting of MI might be warranted in future trials that also carefully assess thrombotic events.
COMPETENCY IN MEDICAL KNOWLEDGE: Pathological left ventricular remodeling after myocardial infarction is associated with thinning of the infarct zone and can lead to ventricular dilation and heart failure. Intracoronary injection of a bioabsorbable cardiac matrix material 2 to 5 days after primary reperfusion in patients with STEMI, however, did not attenuate left ventricular dilation or adverse clinical events within the following 6 months.
TRANSLATIONAL OUTLOOK: Further studies of bioabsorbable cardiac matrix during primary PCI should address its impact on functional outcomes such as exercise capacity, as well as on thromboembolic events.
For an expanded Methods section and tables, please see the online version of this article.
The PRESERVATION I trial was funded by Bellerophon Therapeutics Inc. Three former Bellerophon employees participated as coauthors (Drs. Liu, Levy, and Heyrman). Dr. Rao has received research funding from Bellerophon Therapeutics Inc. Dr. Zeymer has received honoraria for participation at steering committee meetings of PRESERVATION I. Dr. Douglas has received research funding from Bellerophon Therapeutics Inc. Dr. Levy is a contractor for and was a consultant to Bellerophon Therapeutics Inc. during the conduct of this study; and owns stock options in the company. Dr. Gibson has received research funding from Bellerophon Therapeutics Inc. Dr. Tanguay’s institution received payment from the Duke Clinical Research Institute for the PRESERVATION I trial; and he has received funding as a steering committee member; and has received a research grant from Ikaria-Bellerophon. Dr. Kasprzak has received investigator fees from Bellerophon Therapeutics Inc. Dr. Henry is a steering committee member for the PRESERVATION I study. Dr. Chew has received speaker honoraria from Medscape and AstraZeneca. Dr. Lopez-Sendon has received research grants from Daiichi-Sankyo, GSK, Servier, Sanofi, Novartis, the Menarini Group, Pfizer, Merck, and AstraZeneca; and has been an advisor for or received honoraria from Amgen, the Menarini Group, Merck, Servier, Sanofi, Novartis, and AstraZeneca. Dr. Heyrman was an employee of Ikaria-Bellerophon at the time of the study; and owns stock in Bellerophon Therapeutics Inc. Dr. Krucoff has received research grant support from Bellerophon Therapeutics Inc. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- 6-minute walk test
- bioabsorbable cardiac matrix
- extracellular matrix
- heart failure
- Kansas City Cardiomyopathy Questionnaire
- left ventricle/ventricular
- left ventricular end-diastolic volume index
- myocardial infarction
- New York Heart Association
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- Thrombolysi In Myocardial Infarction
- Received April 7, 2016.
- Accepted May 10, 2016.
- American College of Cardiology Foundation
- St John Sutton M.,
- Pfeffer M.A.,
- Plappert T.,
- et al.
- Solomon S.D.,
- Skali H.,
- Anavekar N.S.,
- et al.
- Landa N.,
- Miller L.,
- Feinberg M.S.,
- et al.
- Santos-Gallego C.G.,
- Picatoste B.,
- Njerve I.U.,
- et al.
- Rao S.V.,
- Zeymer U.,
- Douglas P.S.,
- et al.
- al-Shamkhani A.,
- Duncan R.
- Douglas P.S.,
- DeCara J.M.,
- Devereux R.B.,
- et al.
- Frey N.,
- Linke A.,
- Suselbeck T.,
- et al.
- Kramer D.G.,
- Trikalinos T.A.,
- Kent D.M.,
- Antonopoulos G.V.,
- Konstam M.A.,
- Udelson J.E.
- Konstam M.A.,
- Kramer D.G.,
- Patel A.R.,
- Maron M.S.,
- Udelson J.E.