Author + information
- Received June 8, 2017
- Revision received January 18, 2018
- Accepted January 25, 2018
- Published online March 14, 2018.
- Simone Frea, MDa,∗ (, )
- Paolo Centofanti, MDb,
- Stefano Pidello, MDa,
- Francesca Giordana, MDa,
- Virginia Bovolo, MDa,
- Andrea Baronetto, MDb,
- Beatrice Franco, MDa,
- Marco Matteo Cingolani, MDa,
- Matteo Attisani, MDb,
- Mara Morello, MDa,
- Serena Bergerone, MDa,
- Mauro Rinaldi, MDb and
- Fiorenzo Gaita, MDa
- aDivision of Cardiology, Città della Salute e della Scienza University Hospital of Torino, Torino, Italy
- bDivision of Cardiac Surgery, Città della Salute e della Scienza University Hospital of Torino, Torino, Italy
- ↵∗Address for correspondence:
Dr. Simone Frea, Division of Cardiology, Cardiovascular and Thoracic Department, Città della Salute e della Scienza Hospital and University of Turin, Corso Bramante 88, 10126 Torino, Italy.
Objectives The aim of this prospective study was to validate an echocardiographic protocol derived from 5 HeartWare left ventricular assist device (HVAD) patients for the noninvasive evaluation of right atrial pressure (RAP) and left atrial pressure (LAP) in HVAD patients.
Background Echocardiography is an invaluable tool to optimize medical treatment and pump settings and also for troubleshooting residual heart failure. Little is known about the echocardiographic evaluation of hemodynamic status in HVAD patients.
Methods Right heart catheterization and Doppler echocardiography were performed in 35 HVAD patients. Echocardiography-estimated RAP (eRAP) was assessed using inferior vena cava diameter, hepatic venous flow analysis, and tricuspid E/e′ ratio. Echocardiography-estimated LAP was assessed using E/A ratio, mitral E/e′ ratio, and deceleration time.
Results eRAP and estimated LAP significantly correlated with invasive RAP and LAP (respectively, r = 0.839, p < 0.001, and r = 0.889, p < 0.001) and accurately detected high RAP and high LAP (respectively, area under the curve 0.94, p < 0.001, and area under the curve 0.91, p < 0.001). High eRAP was associated with high LAP (area under the curve 0.92, p < 0.001) and correlated with death or hospitalization at 180 days (odds ratio: 8.2; 95% confidence interval: 1.1 to 21.0; p = 0.04). According to estimated LAP and eRAP, patients were categorized into 4 hemodynamic profiles. Fifteen patients (43%) showed the optimal unloading profile (normal eRAP and normal wedge pressure). This profile showed a trend toward a lower risk for adverse cardiac events at follow-up (odds ratio: 0.2; 95% confidence interval: 0.1 to 1.0; p = 0.05) compared with other hemodynamic profiles.
Conclusions Doppler echocardiography accurately estimated hemodynamic status in HVAD patients. This algorithm reliably detected high RAP and LAP. Notably, high RAP was associated with high wedge pressure and adverse outcome. The benefit of noninvasive estimation of hemodynamic status in the clinical management of patients with left ventricular assist devices needs further evaluation.
Left ventricular assist devices (LVADs) are intended to improve cardiac output and unload the left ventricle without excessive overload on the right cardiac chambers (1–3). Inefficient unloading of the left ventricle may be associated with failure to diminish heart failure symptoms, and right ventricular (RV) performance could be the limiting factor on total cardiac output after LVAD implantation. Echocardiography is an invaluable tool to optimize medical treatment and pump settings and also for troubleshooting possible device malfunctions (4–9). Echocardiographic protocols were prospectively validated in HeartMate II (Thoratec Corporation, Pleasanton, California) LVAD patients (8–11). However, little is known about echocardiographic evaluation in those with HeartWare (Framingham, Massachusetts) LVADs (HVADs). Therefore, we performed a prospective study to validate a pre-specified echocardiographic protocol for the noninvasive evaluation of hemodynamic status in a selected population of HVAD-implanted patients.
Patient population and study design
In this single-center prospective study, all consecutive HVAD patients who underwent right heart catheterization (RHC) between July 2014 and April 2017 were enrolled. Patients underwent RHC to assess or maintain heart transplantation candidacy or for persistent heart failure (12). If no pulse or narrow pulse was present, Doppler blood pressure was measured (13), while pulsatile mean arterial pressure was derived from systolic and diastolic blood pressure. A Doppler blood pressure or mean arterial pressure goal of ≤80 mm Hg was used.
Echocardiographic evaluation was performed by blinded operators (S.F. and M.M.) according to a pre-specified protocol within 60 min before RHC. Follow-up was performed 180 days after RHC.
Adverse outcomes were considered a composite of cardiac death, hospitalization for heart failure, RV mechanical support, or urgent heart transplantation within 180 days of RHC.
The study was drafted according to the Declaration of Helsinki, International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use good clinical practice, and regulatory requirements and was approved by the local Institutional Review Board; patients gave written informed consent.
Two-dimensional transthoracic echocardiography was performed according to a pre-specified protocol using a Philips i33 machine (Philips Medical Systems, Andover, Massachusetts).
Left ventricular (LV) chamber size and function were measured (14). Inspection and qualitative description of inlet cannula was performed. Aortic regurgitation and mitral regurgitation (MR) were assessed using color Doppler (15). Interventricular septal position and aortic valve opening (16) were evaluated. Right or left shift of the septum was respectively considered a marker of inefficient or excessive unloading (8).
Basal RV end-diastolic diameter, tricuspid annular plane systolic excursion, 2-dimensional RV fractional area change, and tissue Doppler–derived tricuspid lateral annular peak systolic velocity (S′) were measured (17). Tricuspid regurgitation was qualitatively assessed. Transtricuspid systolic gradient was estimated by tricuspid regurgitation peak velocity, and RV contraction-pressure index was derived as tricuspid annular plane systolic excursion × transtricuspid systolic gradient (18). Finally we estimated the pulmonary artery pulsatility index (PAPi) (19) as follows: estimated PAPi = transtricuspid systolic gradient/estimated right atrial pressure (eRAP) (see the following text).
Noninvasive hemodynamic protocol
The noninvasive hemodynamic protocol was developed according to the best Doppler echocardiographic knowledge and was previously tested and optimized on a derivation cohort of 5 HVAD patients (patients not included in the study). In this cohort, Doppler echocardiographic estimation of right atrial pressure (RAP) significantly correlated with RAP obtained by RHC (r = 0.915, R2 = 0.837, p = 0.029). Estimation of left atrial pressure (LAP) according to 2 different models (eLAP1 and eLAP2; see below) showed a trend toward a significant correlation with wedge pressure (WP) (respectively, r = 0.830, R2 = 0.69, p = 0.16, and r = 0.794, R2 = 0.630, p = 0.11).
Expiratory and inspiratory inferior vena cava (IVC) diameters within 2 cm from the right atrium and hepatic vein flow were measured in subcostal views, while peak early transtricuspid inflow velocity and tissue Doppler analysis of tricuspid annular velocities were measured in a right ventricle–focused apical 4-chamber view.
RAP was estimated using IVC diameter and collapse (20), hepatic venous flow pattern (hepatic venous systolic-to-diastolic wave ratio) (21), and hepatic venous systolic filling fraction (ratio of systolic and diastolic hepatic venous velocity-time integrals (VTIs) (systolic VTI/[systolic VTI + diastolic VTI]) (22) and using the tricuspid E/e′ ratio (23,24) (Figures 1A and 2A to 2D⇓⇓). IVC diameter >2.1 cm, hepatic venous systolic-to-diastolic wave ratio <1, hepatic venous systolic filling fraction < 55%, and tricuspid E/e′ ratio >6 were considered markers of high (>10 mm Hg) RAP. Lower and higher cutoff values of eRAP were derived from the models as follows: RAP = 21.6 − 24 × hepatic venous systolic filling fraction for hepatic vein flow analysis and RAP = 1.62 × E/e′ + 2.13 for tricuspid E/e′ ratio analysis (22,24). Finally, the conclusive eRAP was the average of RAP values estimated by at least 2 of the 3 aforementioned parameters.
The first estimation of LAP (eLAP1) was derived from eRAP and interatrial septal position. Interatrial septal position was assessed at diastole using the left parasternal short-axis view and/or 4-chamber view (8,25). eLAP1 was considered equal to eRAP if the interatrial septum position was neutral, while eLAP1 was 5 mm Hg higher or lower if the septum was deviated respectively to the right or to the left side.
The second multiparametric model (eLAP2) used transmitral Doppler analysis (diastolic pattern analysis assessed by E/A ratio and deceleration time [DT] of the E-wave and the mitral deceleration index, as the DT/E-wave peak velocity ratio) (26), tissue Doppler analysis of mitral annular velocities (mitral septal E/e′ ratio), and MR degree, as a direct marker of LV load (Figure 1B). Tricuspid regurgitation peak velocity was not included in the eLAP2 model, because it was independently evaluated as an indirect measure of LAP (27). Pre-specified predictors of high (>15 mm Hg) WP were a restrictive filling pattern (E/A ≥2 and DT <160 ms), a mitral deceleration index <2 ms/(cm/s), a septal E/e′ ratio ≥15 (28), MR ≥3+/4+, or diastolic MR and a tricuspid regurgitation peak velocity >2.8 m/s. Finally, the conclusive eLAP2 value was the average of the eLAP values estimated by the 4 (or fewer if 1 or more parameters were not available) aforementioned parameters.
A third model for the detection of high LAP, proposed by Estep et al. (9) and including E/A ratio, RAP assessed by IVC diameter and hepatic venous flow, systolic pulmonary artery pressure (sPAP), E/e′ ratio, and left atrial volume index, was used.
Finally, according to eLAP2 and eRAP, patients were classified into 4 different hemodynamic profiles: optimal unloading (normal eLAP and eRAP), RV failure (normal eLAP and high eRAP), LV failure (high eLAP and normal eRAP), and biventricular failure (high eLAP and high eRAP).
RHC was performed in our catheterization laboratory by an operator blinded to all Doppler echocardiographic data. RAP, pulmonary artery pressure, and mean WP were measured; cardiac output was derived. Mean WP was considered high when >15 mm Hg and RAP when >10 mm Hg. To evaluate RV function, we calculated RV stroke work index, PAPi ([sPAP − diastolic pulmonary artery pressure]/mean RAP), and the mean RAP/mean WP ratio. To assess RV load, we evaluated pulmonary vascular resistance and the effective arterial elastance (29) as sPAP/SVI.
Continuous variables are expressed as mean ± SD and were compared using analysis of variance. Categorical variables are presented as counts and percentages and were compared using the chi-square test. The correlations between variables were evaluated using the Pearson or Spearman rho test and were graphically appraised according to Bland-Altman methods. The same tests were used to evaluate interobserver and intraobserver variability.
Correlations between variables and high LAP or RAP were tested in cross tabulation tables using the Fisher exact test and the Student t test for categorical and continuous variables, respectively.
Receiver-operating characteristic curves were produced to test the abilities of the variables to predict high left and right filling pressures.
Kaplan-Meier curves were used to measure freedom from adverse clinical events according to different noninvasive hemodynamic profiles.
A 2-sided p value <0.05 was considered to indicate statistical significance. All analyses were performed using SPSS version 20.0 (IBM, Armonk, New York).
Thirty-five HVAD patients (mean age 56.9 ± 10.5 years, mean 16 ± 12 months with HVAD support [range 3 to 48 months]) who underwent RHC (24 patients for heart transplantation candidacy and 11 for heart failure) were consecutively enrolled. At HVAD implantation 30 patients had an INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) profile ≤3. Table 1 summarizes the characteristics of the population at RHC. Most patients (n = 23 [66%]) were implanted with a “bridge to transplantation” or “bridge to candidacy” indication.
At 180 days from RHC, 8 patients (23%) experienced adverse outcomes (7 were hospitalized for heart failure, and 1 died of RV failure). Six patients (17%) had clinically relevant suction episodes. Two patients needed aortic valve surgery because of severe aortic regurgitation with cardiogenic shock. Four patients (11%) underwent elective heart transplantation during overall follow-up.
The enrolled population showed a good range of values at RHC (WP, median 15.5 mm Hg [range 4 to 34 mm Hg]; RAP, median 10 mm Hg [range 3 to 24 mm Hg]). Main hemodynamic characteristics are listed in Table 1. Sixteen patients (46%) showed high WP.
Doppler echocardiographic findings
Main Doppler echocardiographic data are summarized in Table 1. Two patients showed severe aortic regurgitation.
RAP estimation and detection of elevated RAP
Average eRAP was 11.1 ± 4.3 mm Hg (median 12 mm Hg; range 3 to 18 mm Hg). RAP estimated by IVC, tricuspid E/e′ ratio, and hepatic vein hepatic venous systolic-to-diastolic wave ratio correlated significantly with invasive RAP (Table 2). Hepatic vein flow analysis was the most accurate. Multiparametric eRAP showed the highest correlation with invasive RAP (r = 0.839, R2 = 0.704; mean difference 0.8 ± 3.0, Student t test for mean difference different from zero, p = 0.12) (Figures 3A and 3C) and the greatest diagnostic accuracy (area under the curve [AUC]: 0.94, p < 0.001).
LAP estimation and detection of elevated LAP
Patients with high WP showed lower mitral deceleration index (1.71 ± 0.57 ms/[cm/s] vs. 3.67 ± 1.65 ms/[cm/s]; p < 0.01), higher septal E/e′ ratio (18.5 ± 4.4 vs. 10.2 ± 4.0; p < 0.001), and shorter DT (125.1 ± 33.4 ms vs. 169.3 ± 75.6 ms; p = 0.04), while left atrial volume index showed no difference. No patient in the high WP group showed interventricular septum deviated to left, while 5 in the normal WP group did (p < 0.05), 4 of whom underwent clinically relevant suction episodes during follow-up (odds ratio: 38; p < 0.01). No significant difference in aortic valve opening status was found. The eLAP1 model showed a good correlation with WP (Table 3). This result was driven by eRAP. The eLAP2 model correlated well with high WP (AUC: 0.90; p < 0.001), though it underestimated WP higher than 20 mm Hg (mean difference 3.7 ± 3.6, Student t test for mean difference different from zero; p < 0.01) (Figures 3B and 3D). The eLAP model of Estep et al. (9) was also a fair predictor of high WP in our setting of patients (AUC: 0.73; p = 0.04). However eLAP according to Estep et al. showed lower specificity. This result was driven mainly by left atrial volume index.
Finally, we performed a post hoc analysis. We modified the algorithm proposed by Estep et al. (9) to elaborate a more accurate and practice algorithm (the “Estep modified” model). It was a 2-step algorithm. Multiparametric estimation of RAP was the first step. Afterward, eLAP was assessed using eRAP, E/A, and E/e′ (Figure 4). This brief model allowed a noninvasive estimation of hemodynamic status in all patients with excellent accuracy (AUC for high WP: 0.97).
Doppler echocardiographic estimation of pulmonary hemodynamic status and RV function
There was a good correlation between Doppler echocardiographic estimation of sPAP and sPAP on RHC (r = 0.728, p < 0.001). Among noninvasive parameters of RV function, Doppler echocardiographic PAPi and eRAP/eLAP2 correlated significantly with invasive PAPi and RAP/WP (respectively, r = 0.581, p = 0.002, and r = 0.488, p = 0.01).
Noninvasive hemodynamic profiles
According to eLAP and eRAP, patients were eventually categorized into hemodynamic profiles. Clinical, Doppler echocardiographic, and invasive data according to hemodynamic profiles are shown in Table 4. Fifteen patients (43%) showed the optimal unloading profile, 14 patients (40%) the biventricular failure profile, 1 the LV failure profile, and 5 (14%) the RV failure profile. Compared with the optimal unloading group, the other 2 groups showed worse RV function on RHC, as indicated by PAPi and the mean RAP/mean WP ratio. In particular, the RV failure group showed low-load RV dysfunction (lower sPAP, RV–right atrial gradient, pulmonary vascular resistance, and arterial elastance), while the biventricular failure group showed high-load RV dysfunction (higher sPAP, RV–right atrial gradient, pulmonary vascular resistance, and arterial elastance). Notably, patients with the optimal unloading profile showed a lower risk for adverse cardiac events at follow-up (odds ratio: 0.2; 95% confidence interval: 0.1 to 1.0; p = 0.05) (Figure 5) compared with other profiles. In contrast, high eRAP significantly predicted adverse outcomes (odds ratio: 8.2; 95% confidence interval: 1.1 to 21.0; p = 0.04).
Reproducibility of echocardiographic parameters
The reproducibility of Doppler echocardiographic and tissue Doppler imaging measurements in our laboratory was previously reported (30). Intraobserver (eRAP: r = 0.999, p = 0.001; eLAP2: r = 0.986, p = 0.001) and interobserver (eRAP: r = 0.994, p = 0.001; eLAP2: r = 0.931, p = 0.001) estimation of eRAP and eLAP2 showed a very good agreement between measurements.
This is the first prospective study to report on the noninvasive Doppler echocardiographic evaluation of hemodynamic status in HVAD patients. We validated a pre-specified protocol for the noninvasive detection of high RAP and LAP.
Estimation of RAP and LAP
The good correlation between IVC and RAP was comparable with that observed in patients without LVADs. Nevertheless, though very specific, a dilated IVC showed low sensitivity in the detection of high RAP. Previous reports suggested that IVC should not be used alone for a reliable estimation of RAP because of its variability and overlap between patients with normal RAP and those with mildly elevated RAP (31,32). In this respect, hepatic vein flow analysis showed very good sensitivity and diagnostic accuracy, as previously reported (9). The multiparametric approach showed the best accuracy, as it increased sensitivity without affecting specificity.
Despite inlet cannula artifacts, using off-axis views, an adequate mitral pulsed and tissue Doppler analysis of diastolic pattern was reliably obtained in the majority of patients. As previously reported by Estep et al. (9) in HeartMate II–assisted patients, pulsed and tissue Doppler showed a fair linear correlation with WP, and a multiparametric evaluation showed better diagnostic accuracy than single parameters. This suggests that the noninvasive assessment of hemodynamic status is accurate and reproducible and that it is probably not affected by the type of continuous-flow LVAD.
We believe that our study, in addition to that of Estep et al. (9), confirms that noninvasive estimation of hemodynamic status in LVAD patients is accurate and reproducible. In fact, most of our results were similar to those of Estep et al. (9). However, the eLAP2 model showed a better accuracy than that of Estep et al. This was due to the low diagnostic accuracy of left atrial volume. In fact, in our cohort, almost all patients had severely dilated left atria independent of WP, and this led to an overestimation of high WP. In addition, our results confirmed the strong association between high eRAP and high WP found by Estep et al. (9), as the majority of patients with high WP showed high eRAP. This probably depends on a worsening of RV adaptation to load with LVAD support, as suggested Houston et al. (33).
Moreover, high eRAP was the strongest predictor of adverse outcomes. In this respect, the estimation of RAP should be the first step in the evaluation because of its prognostic value and its diagnostic accuracy. We therefore developed a post hoc 2-step simplified model (the Estep modified model) whose first step is the multiparametric assessment of eRAP. Afterward, WP is assessed. This easier approach focuses on the interplay between the LVAD and the right ventricle (see the following text).
In line with previous studies, RV function was not captured by conventional Doppler echocardiographic parameters (tricuspid annular plane systolic excursion, fractional area change, and S′ peak velocity), while direct or indirect RV hemodynamic evaluation (PAPi, RV stroke work index, RV contraction-pressure index, RV–right atrial gradient) did. In this regard, the RV failure subgroup showed primary severe RV dysfunction, while in the biventricular failure subgroup, RV dysfunction seemed to be a consequence of higher RV afterload.
Doppler echocardiographic profiles and clinical implications
The hemodynamic profiles by eRAP and eLAP outlined some differences in terms of prognosis, RV function, and treatment (Figure 6). Patients with the optimal unloading profile had better outcomes, while those with high eRAP (both RV failure and biventricular failure profiles) had a higher risk for heart failure.
This classification may help clinicians in therapeutic and device management. In case of high WP, hypertension and other specific causes of high LV afterload (e.g., LVAD thrombosis, aortic insufficiency, inadequate unloading) should be considered. An increase in LVAD speed rate may be considered a therapeutic option once other causes are excluded. In case of RV dysfunction, the use of diuretic agents and inotropes may be considered. However, in case of overt primary RV failure, it could be necessary to reduce the LVAD speed and hence to accept suboptimal LV unloading to achieve outpatient stable RV compensation.
The main limits of the study are the monocentric design and the small number of patients enrolled. Even though we collected a wide spectrum of WP and RAP values, an external validation of our algorithm is needed. Also the eLAP Estep modified model, which resulted from a post hoc analysis, lacks a validation cohort.
We collected satisfactory Doppler echocardiographic signals in many patients, but this required the use of off-axis and sometimes atypical sample volume positions and high angles of insonation. Besides, noninvasive estimation was performed by a team with expertise in diastolic evaluation. This could have affected the algorithm accuracy and reproducibility, as Doppler echocardiography is an operator-dependent technique.
Moreover, this study involved only HVAD patients, so the algorithm could not fit other types of LVADs.
The clinical role of noninvasive hemodynamic profiles was not the main goal of the study, and it may have been underpowered to adequately evaluate the association of profiles with different prognosis and treatment. In particular, only 1 patient fit the LV failure profile. The selection of patients implanted (INTERMACS ≤4, mildly dysfunctional right ventricles) may have played a role.
Noninvasive evaluation of hemodynamic status in HVAD patients by Doppler echocardiography is feasible and reproducible. Both RAP and WP were accurately estimated using a multiparametric approach. High RAP is associated with high WP and adverse outcomes. The benefit of noninvasive evaluation of hemodynamic status in the clinical assessment of LVAD patients should be evaluated.
COMPETENCY IN MEDICAL KNOWLEDGE: In LVAD patients, noninvasive estimation of high LAP and RAP by Doppler echocardiography is accurate and reproducible and identifies patients with different clinical presentation and prognosis. Furthermore, high RAP was associated with inadequate unloading of the left ventricle and was the strongest predictor of adverse cardiac outcomes.
TRANSLATIONAL OUTLOOK: More studies are needed to assess whether the addition of noninvasive evaluation of hemodynamic status can guide the management of LVAD patients, whether it improves the achievement of optimal unloading of both ventricles, and whether it influences survival and the quality of life of LVAD patients.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- area under the curve
- deceleration time
- estimated left atrial pressure
- estimated right atrial pressure
- HeartWare left ventricular assist device
- inferior vena cava
- left atrial pressure
- left ventricular
- left ventricular assist device
- mitral regurgitation
- pulmonary artery pulsatility index
- right heart catheterization
- right ventricular
- right atrial pressure
- systolic pulmonary artery pressure
- hepatic venous velocity-time integral
- wedge pressure
- Received June 8, 2017.
- Revision received January 18, 2018.
- Accepted January 25, 2018.
- 2018 American College of Cardiology Foundation
- Jorde U.P.,
- Kushwaha S.S.,
- Tatooles A.J.,
- et al.,
- for the HeartMate II Clinical Investigators
- Slaughter M.S.,
- Pagani F.D.,
- McGee E.C.,
- et al.,
- for the HeartWare Bridge to Transplant ADVANCE Trial Investigators
- Horton S.C.,
- Khodaverdian R.,
- Chatelain P.,
- et al.
- Estep J.D.,
- Stainback R.F.,
- Little S.H.,
- Torre G.,
- Zoghbi W.A.D.
- Topilsky Y.,
- Hasin T.,
- Oh J.K.,
- et al.
- Estep J.D.,
- Vivo R.P.,
- Krim S.R.,
- et al.
- Uriel N.,
- Morrison K.A.,
- Garan A.R.,
- et al.
- Feldman D.,
- Pamboukian S.V.,
- Teuteberg J.J.,
- et al.,
- for the International Society for Heart and Lung Transplantation
- Lanier G.M.,
- Orlanes K.,
- Hayashi Y.,
- et al.
- Zoghbi W.A.,
- Enriquez-Sarano M.,
- Foster E.,
- et al.
- Rudski L.G.,
- Lai W.W.,
- Afilalo J.,
- et al.
- Kang G.,
- Ha R.,
- Banerjee D.
- Nagueh S.F.,
- Kopelen H.A.,
- Zoghbi W.A.
- Sade L.E.,
- Gulmez O.,
- Eroglu S.,
- Sezgin A.,
- Muderrisoglu H.
- Tei C.,
- Tanaka H.,
- Kashima T.,
- Yoshimura H.,
- Minagoe S.,
- Kanehisa T.
- Nagueh S.F.,
- Smiseth O.A.,
- Appleton C.P.,
- et al.
- Frea S.,
- Pidello S.,
- Bovolo V.,
- et al.
- Mintz G.S.,
- Kotler M.N.,
- Parry W.R.,
- Iskandrian A.S.,
- Kane S.A.
- Houston B.A.,
- Kalathiya R.J.,
- Hsu S.,
- et al.