Author + information
- Received June 13, 2017
- Revision received July 28, 2017
- Accepted August 1, 2017
- Published online September 4, 2017.
- Brian A. Boe, MDa,∗ (, )
- Jeffrey D. Zampi, MDb,
- Kevin F. Kennedy, MSc,
- Natalie Jayaram, MD, MSBd,
- Diego Porras, MDe,
- Susan R. Foerster, MDf and
- Aimee K. Armstrong, MDa
- aThe Heart Center, Nationwide Children’s Hospital, Columbus, Ohio
- bDivision of Pediatric Cardiology, University of Michigan, Ann Arbor, Michigan
- cCardiovascular Research, Saint Luke’s Mid America Heart Institute, Kansas City, Missouri
- dDepartment of Cardiology, Children’s Mercy Kansas City, Kansas City, Missouri
- eDepartment of Cardiology, Boston Children’s Hospital, Boston, Massachusetts
- fDivision of Pediatric Cardiology, Children’s Hospital of Wisconsin, Milwaukee, Wisconsin
- ↵∗Address for correspondence:
Dr. Brian A. Boe, The Heart Center, Nationwide Children’s Hospital, 700 Children’s Drive, Columbus, Ohio 43205-2664.
Objectives The aim of this study was to evaluate practice patterns and outcomes of a contemporary group of patients undergoing balloon aortic valvuloplasty (BAV) for congenital aortic stenosis (AS).
Background BAV is the most common treatment for isolated congenital AS.
Methods Within the IMPACT (Improving Pediatric and Adult Congenital Treatments) Registry, all BAV procedures performed between January 2011 and March 2015 were identified. Procedures were separated into those performed for critical versus noncritical AS. Outcomes were stratified into optimal, adequate, and inadequate, with optimal and adequate outcomes defining “successful” procedures. Multivariate logistic regression was used to identify patient and procedural characteristics associated with unsuccessful BAV. Mortality and adverse events rates were compared across patient cohorts.
Results Of the 1,026 isolated BAV procedures captured in IMPACT, 718 (70%) were “successful.” Success rates were 70.9% for noncritical AS (n = 916) and 62.7% for critical AS (n = 110). Multivariate analysis revealed that prior cardiac catheterization, mixed valve disease, baseline aortic valve gradient >60 mm Hg, baseline aortic insufficiency greater than mild, presence of a trainee, and multiple balloon inflations were associated with unsuccessful BAV in the noncritical AS cohort. There were no factors associated with unsuccessful procedures in the critical AS group. No procedural deaths occurred, but 2.4% of patients did not survive to hospital discharge. Adverse events occurred in 15.8% of all cases and were more frequent in procedures performed for critical AS (30.0% vs. 14.1%; p < 0.001).
Conclusions BAV is an effective treatment for congenital AS with low rates of mortality and adverse events. Patients with critical AS have a higher risk for procedure-related adverse events.
Transcatheter balloon aortic valvuloplasty (BAV) is considered the first-line palliative therapy for congenital aortic stenosis (AS) at most centers. The procedure has been shown to avoid or delay aortic valve surgery in long-term follow-up studies (1–5). Technical factors of the procedure and associated outcomes have been studied since the inception of BAV in the 1980s. These early studies identified balloon–to–aortic annulus ratio >1, younger age, and unicuspid or thickened valve morphology as risk factors for worse outcomes (6–11). Additionally, higher rates of mortality and complications were documented in procedures performed for neonatal “critical” AS (12,13).
Despite several early studies documenting the risk factors and outcomes for BAV, current procedural outcomes are less understood. Although 2 recent studies reported on the acute success of BAV, these studies were somewhat limited in their sample sizes (14,15). Given the significant improvement in catheter-based equipment (e.g., low-profile balloons, smaller sheaths) and techniques (e.g., rapid ventricular pacing, balloon–to–aortic valve annulus ratio <1) over the past few decades (14,16–18), an understanding of contemporary practice patterns and outcomes is important.
Since 2011, the IMPACT Registry has been collecting multicenter prospective data on congenital cardiac catheterizations, including BAV procedures. This registry contains the largest available dataset of patients undergoing catheterization for congenital heart disease and presents a unique opportunity to study outcomes of patients undergoing BAV. The aim of this study was to describe practices and acute outcomes of BAV in the current era, using the IMPACT Registry. In addition, we sought to evaluate factors associated with acute outcomes and complications.
The IMPACT Registry, part of the National Cardiovascular Data Registry, is an initiative of the American College of Cardiology Foundation and is focused on patients with congenital heart disease who undergo cardiac catheterization (19). Demographic, clinical, procedural, and institutional data variables are voluntarily entered by participating centers using a Web-based portal, and data are then collected in a secure, centralized database. Data entered in version 1.0.1 of the IMPACT Registry were used for this study. A complete description of data variables captured by the IMPACT Registry version 1.0.1 can be found at https://www.ncdr.com/WebNCDR/docs/default-source/public-data-collection-documents/impact_v1_datacollectionform_1-0-1.pdf?sfvrsn=2. This study was performed using deidentified data from the IMPACT Registry and thus meets criteria for research not requiring specific informed consent. The collected data are subject to rigorous quality assurance standards set by the Data Quality Program for all National Cardiovascular Data Registry registries, and auditing procedures were developed during the study time frame (20,21).
Standardized and pre-specified procedural information collected for BAV procedures within the IMPACT Registry were included in the study. The technical aspects of BAV have been previously described (8,9,12). Each procedure was performed at the discretion of the operator. Echocardiographic imaging is most commonly used to evaluate AS in the clinical setting, and BAV is performed under fluoroscopic guidance with angiography used to evaluate immediate outcomes (22). Data were collected from January 2011 through March 2015. Only data meeting pre-specified criteria by the National Cardiovascular Data Registry for completeness and accuracy are included in analytic datasets and used for quality reporting back to participating sites (19). Procedures with missing data fields for study outcomes were excluded from analysis. In an effort to describe the outcomes attributed to BAV, patients were excluded if another catheter-based intervention was performed in concert. Patients were also excluded from the study if they were diagnosed with single-ventricle disease or were receiving extracorporeal membrane oxygenation support at the start of the procedure.
The patient population was subdivided by diagnosis into either critical AS or noncritical AS. Critical AS was defined by patients who were <1 month of age at the time of the procedure and receiving prostaglandin infusion within 1 day of the catheterization date. Procedural data for BAV collected by the IMPACT Registry include the catheter-measured peak systolic ejection gradient (PSEG) across the aortic valve and the degree of angiographic aortic insufficiency (AI) as graded by the operator (14,15,23,24):
• 1+ (mild), a small amount of contrast enters the left ventricle in diastole
• 2+ (moderate), faint opacification of the entire chamber occurs
• 3+ (moderately severe), the left ventricular chamber is well opacified and equal in density with the ascending aorta
• 4+ (severe), characterized by complete, dense opacification of the left ventricle chamber in 1 beat, and the left ventricle appears more densely opacified than the ascending aorta.
Baseline procedural data were used to classify patients as having either isolated AS or mixed disease. Isolated AS was defined as patients with AS and no AI. Patients with mixed disease had grades 1+ to 4+ AI, in addition to AS.
The primary outcomes of interest were procedural success and the occurrence of adverse events. Procedural success was defined by classifying outcomes into 3 groups (optimal, adequate, and inadequate) on the basis of the final PSEG and AI following the procedure, as previously published (14,15). These definitions are based on a combination of empirical data and expert consensus (2,14,15):
• Optimal outcome, PSEG ≤35 mm Hg and no AI
• Adequate outcome, PSEG ≤35 mm Hg and 1+ (mild) AI for isolated AS; PSEG ≤35 mm Hg and no worsening AI for patients with mixed disease
• Inadequate outcome: PSEG >35 mm Hg and/or >1+ (mild) AI for isolated AS; PSEG >35 mm Hg or worse AI compared with pre-procedural AI for patients with mixed disease.
A successful BAV included both optimal and adequate outcomes for the purposes of analysis.
The IMPACT Registry collects the residual aortic valve gradient and degree of AI following each balloon inflation. Post–balloon inflation data are reported at the discretion of the participating center either following inflation of every balloon (i.e., the same balloon inflated twice at the same or different pressures), inflation of larger balloons, or a combination of the two.
Adverse events are reported as captured by the IMPACT Registry and dichotomized into minor and major according to previously published models (25,26). Mortality was defined as any death recorded in the IMPACT Registry before hospital discharge.
Patient and procedural factors
A number of patient and procedural factors were evaluated for associations with procedural outcomes. Patient factors included age, sex, race, weight, prematurity (<37 weeks’ gestational age), genetic conditions, hospital status, cardiac diagnosis, noncardiac diagnosis, and valve morphology (unicuspid, bicuspid, tricuspid, quadricuspid, or uncertain). Procedural factors included procedural status (elective, urgent, emergency, or salvage), type of procedural sedation, access location, balloon technique (single vs. double), use of balloon stabilization techniques (e.g., rapid ventricular pacing), and balloon–to–aortic valve annulus ratio. The effective balloon diameter of a double-balloon technique is calculated by dividing the sum of the balloons’ diameters by 1.3.
Study factors and outcomes were described in the overall cohort using standard summary statistics. Nominal and categorical data are reported as number and percentage of the total. Interval data are reported as mean ± SD or median (interquartile range [IQR]) as appropriate. Characteristics of successful and unsuccessful BAV were compared using the Student t test for continuous variables and the chi-square or Fisher exact test for categorical variables. Factors with p values <0.05 were considered significant. Significant variables identified in univariate analysis and previous studies were used to create a multivariate model. Multivariate analysis via logistic regression was performed on the noncritical AS cohort. The underpowered critical AS group was evaluated via multivariate analysis of all procedures with interaction testing of the patient cohorts. Post–balloon inflation data (residual aortic valve gradient and degree of AI) were separated by the total number of balloon inflations and compared.
Patient and procedural characteristics
A total of 1,126 BAV procedures were recorded within the IMPACT Registry from January 2011 through March 2015. Figure 1 summarizes the study cohort by diagnosis and outcomes. A total of 100 procedures were excluded from analysis. Missing outcome data (i.e., final aortic valve gradient and/or AI) led to the exclusion of 29 procedures. Patient demographic information from the 1,026 BAV procedures included in analysis is listed in Table 1. Neonates (patients <1 month of age) were the most prevalent age group (27.4%), followed by infants (1 to 11 months; 25.2%), and adolescents (11 to 17 years; 18.6%). The majority of patients had noncritical AS (n = 916 [89.3%]). Procedural status was reported as elective in 71.8% of cases, urgent in 21.8%, an emergency in 6.2%, and a salvage procedure in 0.3%. The majority of BAV procedures (n = 706 [68.8%]) were performed in an outpatient setting. Following the procedure, 54% of outpatient cases were observed for 23 h, 22.8% were discharged home the same day, 12.9% were admitted to the inpatient unit, and 10.3% were admitted to the intensive care unit. Inpatients who underwent BAV returned to the intensive care unit following 79.4% of the cases and to the inpatient floor following 20.6% of the cases.
Indications for BAV included AS gradient (81.3%), left ventricular dysfunction (12.3%), symptoms (4.8%), and abnormal stress test or electrocardiographic findings (1.6%). The vast majority of procedures were performed under general anesthesia (92.8%), with the patient intubated (90.9%) and with an anesthesiologist present (95.8%). Arterial and venous access was obtained in 90.5% of cases. Isolated arterial access was performed in a small percentage of patients (8.2%), and isolated venous access was used in 1.3%. The most common arterial access site was the femoral artery (93.9%), followed by the carotid artery (4.5%) and umbilical artery (1.1%). Of note, there were 3 reported cases performed via the right radial artery. Systemic heparinization was used in 97.0% of cases.
A wide variety of balloon types were used for BAV. The most common type of balloon was the Tyshak II Percutaneous Transluminal Valvuloplasty catheter (NuMED, Hopkinton, New York), which was used in 52% of cases, followed by the Tyshak Mini Pediatric Valvuloplasty catheter (NuMED) in 12.7% and XXL Balloon Dilation Catheter (Boston Scientific, Marlborough, Massachusetts) in 4.2% of cases. Most operators performed BAV using a single-balloon technique (94.8%), and a double-balloon technique was used in 5.2%. The median effective final balloon diameter–to–aortic valve annulus ratio was 0.94 (IQR: 0.88 to 1.00). A balloon stabilization technique was used in 68.1% of noncritical AS cases and in 48.1% of critical AS cases. The mean fluoroscopy time and contrast volume were 23.3 ± 15.5 min and 53.6 ± 52.8 ml, respectively.
The acute outcomes of BAV for the entire cohort were evenly distributed across the outcome definitions. Optimal, adequate, and inadequate results were seen in 34.5%, 35.5%, and 30.0% of cases, resulting in procedural success rates of 70.0%. Unsuccessful procedures were attributed to significant AI (greater than mild or worse than baseline) in 12.1% of procedures, a significant residual aortic valve gradient (>35 mm Hg) in 11.4% of procedures, and a combination of the two in 6.5% of procedures.
Noncritical AS outcomes
Acute results of BAV procedures are summarized in Table 2. Of the 916 procedures, 34.3% resulted in optimal outcomes, 36.6% adequate outcomes, and 29.1% inadequate outcomes, yielding a procedural success rate of 70.9%. BAV was unsuccessful in patients with noncritical AS secondary to significant residual AS in 11.9% of patients, significant AI in 11.4% of patients, and a combination of the two in 5.8% of patients.
Factors associated with unsuccessful BAV in the noncritical AS group were separated into patient and procedural factors. In univariate analysis, patient factors associated with an unsuccessful procedure included history of cardiac catheterization, mixed disease, higher pre-procedural aortic valve gradient, and the presence of more severe AI before the procedure. Procedural factors included increased number of balloon inflations and the presence of a trainee (Table 2). After multivariate adjustment, the factors significantly associated with an unsuccessful BAV procedure included prior cardiac catheterization, baseline AS gradient >60 mm Hg, mixed aortic valve disease, worse than mild baseline AI, presence of a trainee, and >1 balloon inflation (Table 3).
Critical AS outcomes
Optimal, adequate, and inadequate results were seen in 36.4%, 26.4%, and 37.3% of critical AS cases, with a procedural success rate of 62.7%. Significant AI was the leading cause of unsuccessful procedures in 17.3% of cases, followed by a combination of AS and AI in 12.7% and significant residual AS in 7.3%. There were no significant factors associated with unsuccessful BAV in patients with critical AS (Table 2). A multivariate model with interaction testing of critical versus noncritical AS found no factors associated with unsuccessful BAV (Online Table 1).
Acute outcomes following each balloon inflation
BAV procedures were completed using a single balloon inflation in 58.1% of cases and with multiple balloon inflations in 41.9% of cases. Nearly all procedures performed included ≤4 total balloon inflations (98.6%). The breakdown of the percentage of procedures by the number of balloon inflations is as follows: 1, 58.1%; 2, 27.3%; 3, 10.7%; 4, 2.5%; 5, 0.7%; 6, 0.6%; and 8, 0.1%. Figure 2 shows the change in hemodynamic parameters following each balloon inflation during BAV for patients who underwent 1 to 4 balloon inflations. Subsequent balloon inflations during BAV resulted in a decrease in average aortic valve gradient and were continued until a final acceptable residual gradient was achieved (≤35 mm Hg). In both the noncritical and critical AS groups, there was an increase in the percentage of patients with more than mild AI with the last inflation.
Mortality and adverse events
Table 4 depicts the mortality and adverse events separated by diagnosis. There were no intraprocedural deaths. A total of 25 patients did not survive to hospital discharge. Both in-hospital mortality and 30-day mortality were significantly higher in the critical AS cohort (p < 0.001). Adverse events occurred in 15.8% of all cases. Major adverse events occurred in 11.5% of cases. Procedures performed on patients with critical AS had higher percentages of both total adverse events (30.0% vs. 14.1%; p < 0.001) and major adverse events (27.3% vs. 9.6%; p < 0.001). Significantly higher rates of vascular complications (9.1% vs. 1.2%; p < 0.001), tamponade (2.7% vs. 0.1%; p < 0.001), and subsequent cardiac catheterization (3.6% vs. 0.1%; p < 0.001) were found in patients with critical AS and contributed to the high adverse event rate in this population.
In the current era, isolated congenital valvular AS is most commonly treated via BAV. Over the past 30 years, the BAV procedure has evolved, with the development of new technologies and procedural techniques (5,14,16–18). This multicenter, prospective registry study assessed the current practices and acute outcomes of BAV. In our study, we found that the majority of BAV procedures are effective, with more than two-thirds resulting in procedural “success.” Adverse events, when they occurred, were more likely to occur in those patients undergoing BAV for critical AS. Our study is the largest of its kind and supports the notion that BAV is a relatively safe and effective means of treating patients with AS.
The challenge of BAV lies in accomplishing the 2 primary objectives of the procedure (reducing the aortic valve gradient while limiting the amount of AI), which have been shown to be inversely related to balloon size. This relationship was established decades ago (6–8) and has led to the procedural guidelines used today (27). Despite the lack of formal standardization of technique in the current study, we did find near uniform adherence to this guideline by current operators, who used a median effective final balloon diameter to aortic valve annulus ratio of 0.94 (IQR: 0.88 to 1.00).
Recently, a definition for BAV procedural success was devised by the Boston group (15) and used in a study by Torres et al. (14). The definition accounts for both the post-procedural residual stenosis and insufficiency across the valve. These studies reported BAV success rates of 71% and 82%. Using this definition in the largest congenital study to date (n = 1,026), we found that BAV was successful in 70% of the cases captured within the IMPACT Registry. The majority of unsuccessful procedures were a result of significant or worsening AI. Significant post-procedural AI (worse than mild) was seen in 17.1% of patients, almost identical to the rates reported in contemporary studies (19% and 18%) (14,15).
Previously identified patient risk factors for unsuccessful BAV in the noncritical AS population include older age, history of BAV, and presence of significant baseline AI (14). We found that the following 6 factors were associated with unsuccessful BAV in the noncritical AS cohort: previous cardiac catheterization, mixed valve disease, baseline aortic valve gradient >60 mm Hg, baseline AI greater than mild, presence of a trainee, and multiple balloon inflations. Four of these 6 factors are patient related and support a growing body of research linking BAV outcomes with the intrinsic characteristics of the aortic valve (10,28). The only pure aortic valve characteristic captured by the IMPACT Registry is the leaflet morphology, and this factor was not associated with procedural outcomes in our study. It is of interest that the presence of a trainee was a risk factor for unsuccessful BAV in the noncritical AS group. This is unlikely a true predictor given the discrepant association between patient cohorts. The probable explanation for this association is the high-risk BAV population, for which trainees are present (e.g., academic institutions, nonelective cases). Case complexity was not modeled in this study to evaluate this association further.
Long-term studies have found that significant residual aortic valve gradients and AI are independent risk factors for a poor outcome (2–4). A standardized management plan for BAV was recently published by Porras et al. (15). The group was able to achieve a high procedural success rate by repeating balloon inflations with sequentially larger balloons until it resulted in an acceptable residual gradient and/or unacceptable degree of AI. Our study showed a similar practice pattern, with multiple balloon inflations resulting in a residual aortic valve gradient <35 mm Hg. There was a slight increase in significant (more than mild) AI following the final inflation attributing to the association of multiple balloon inflations with unsuccessful outcomes.
The safety of BAV has significantly improved over our collective experience. In our patient population, consisting of 110 patients with critical AS and 916 patients with noncritical AS, there were no intraprocedural deaths. Hospital mortality occurred in 1.5% of patients with noncritical AS and in 10% of patients with critical AS. Previously reported procedural mortality rates range between 2% and 4% in patients with noncritical AS (3,6–8) and 9% and 14% in patients with critical AS (12,18,29). Many of the early reported mortalities were secondary to anatomic damage, which has improved with miniaturization of interventional technology and improved technique. Historically, adverse events associated with BAV have ranged from 5% to 40% (6,7,12,14,30). This wide range is secondary to a lack of standardization of procedural adverse event terminology, as well as the gradual inclusion of less severe adverse events over time. Adverse events captured in the IMPACT Registry occurred in 15.8% of BAV procedures and were more frequent in patients with critical AS. This higher rate of complications is expected, given the poor left ventricular function and small size of this patient population. Pulse loss has been associated with smaller patients (31) and is captured in the IMPACT Registry under vascular complications if it required intravascular therapy. We found that vascular complications were the most common complication seen in the smaller critical AS cohort.
Regarding surgical repair, several studies have compared long-term outcomes between surgical aortic valvuloplasty and BAV (32,33). These single-center studies have shown that surgical valvuloplasty has a lower rate of reintervention and a modestly lower rate of aortic valve replacement than BAV. However, their total rates of aortic valve replacement following surgical valvotomy are similar to recent intermediate- and long-term BAV outcomes studies (2,3). Without a broader context in a multi-institutional prospective study or even a randomized trial, this issue may be difficult to definitively settle. Regardless, our study was limited to the acute outcomes and could not be compared with long-term surgical data. As both interventions continue to evolve, outcome data must be collected and compared to provide the best patient care.
The limitations of this study need to be considered when interpreting the results. Given the multicenter registry dataset, some of the data may have been previously reported in other studies (4,15,28). This study was limited to the data captured by the IMPACT Registry version 1.0.1. This version of the registry collected data up to hospital discharge following BAV. As a result, only acute outcomes following BAV are reported, and post-procedural adverse events are not captured beyond hospital discharge. This could lead to the underreporting of adverse outcomes. Version 1.0.1 of the registry does not allow attribution of adverse events (i.e., mortality) throughout a hospitalization. It is possible that mortality following BAV was associated with a subsequent surgical procedure (e.g., a patient with critical AS who underwent a BAV followed by a neonatal Ross procedure). As a result, the true etiology of the reported mortality is unknown. Procedural details such as the type of balloon stabilization technique used were not captured. Data entry is not standardized, and variations in data recording across participating centers may affect the outcome analysis. Specifically, the variation of entry for the spectrum of single-ventricle disease and the balloon inflation counter (e.g., recording every balloon inflation vs. recording only the use of balloons of varying diameter or pressure) could introduce bias into the study. The IMPACT Registry reports AI as graded by the operator. This is highly subjective and was not subjected to validation.
BAV provides hemodynamic treatment of both critical and noncritical AS with success rates of 70% across multiple congenital heart centers. Procedure-related mortality and adverse events are low, with patients with critical AS carrying a higher risk.
WHAT IS KNOWN? BAV is the procedure of choice at most centers to treat isolated congenital AS.
WHAT IS NEW? In the largest cohort of patients undergoing BAV for congenital AS to date, this study demonstrates that the procedure is both effective and safe.
WHAT IS NEXT? Continued evaluation and comparison of treatment modalities for AS will further improve patient care.
For a supplemental table, please see the online version of this article.
Dr. Armstrong has received research grants from Medtronic, Edwards Lifesciences, Abbott, PFM Medical, and Gore; is a consultant and proctor for Edwards Lifesciences and Abbott; and is a proctor for B. Braun Interventional Systems. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic insufficiency
- aortic stenosis
- balloon aortic valvuloplasty
- peak systolic ejection gradient
- Received June 13, 2017.
- Revision received July 28, 2017.
- Accepted August 1, 2017.
- 2017 American College of Cardiology Foundation
- Fratz S.,
- Gildein H.P.,
- Balling G.,
- et al.
- Brown D.W.,
- Dipilato A.E.,
- Chong E.C.,
- Lock J.E.,
- McElhinney D.B.
- Sullivan P.M.,
- Rubio A.E.,
- Johnston T.A.,
- Jones T.K.
- Sholler G.F.,
- Keane J.F.,
- Perry S.B.,
- Sanders S.P.,
- Lock J.E.
- Moore P.,
- Egito E.,
- Mowrey H.,
- Perry S.B.,
- Lock J.E.,
- Keane J.F.
- Maskatia S.A.,
- Justino H.,
- Ing F.F.,
- Crystal M.A.,
- Mattamal R.J.,
- Petit C.J.
- Egito E.S.,
- Moore P.,
- O’Sullivan J.,
- et al.
- Bu’Lock F.A.,
- Joffe H.S.,
- Jordan S.C.,
- Martin R.P.
- Magee A.G.,
- Nykanen D.,
- McCrindle B.W.,
- Wax D.,
- Freedom R.M.,
- Benson L.N.
- McElhinney D.B.,
- Lock J.E.,
- Keane J.F.,
- Moran A.M.,
- Colan S.D.
- Messenger J.C.,
- Ho K.K.,
- Young C.H.,
- et al.
- Moore J.W.,
- Vincent R.N.,
- Beekman R.H. III.,
- et al.
- Boe B.A.,
- Norris M.D.,
- Zampi J.D.,
- Rocchini A.P.,
- Ensing G.J.
- Bergersen L.,
- Gauvreau K.,
- Foerster S.R.,
- et al.
- Jayaram N.,
- Beekman R.H. III.,
- Benson L.,
- et al.
- Feltes T.F.,
- Bacha E.,
- Beekman R.H. III.,
- et al.
- Petit C.J.,
- Gao K.,
- Goldstein B.H.,
- et al.
- Siddiqui J.,
- Brizard C.P.,
- Galati J.C.,
- et al.