Author + information
- Received August 5, 2014
- Revision received October 13, 2014
- Accepted October 21, 2014
- Published online February 10, 2015.
- Raquel Yotti, MD, PhD∗∗ (, )
- Javier Bermejo, MD, PhD∗,
- Enrique Gutiérrez-Ibañes, MD∗,
- Candelas Pérez del Villar, MD∗,
- Teresa Mombiela, MD∗,
- Jaime Elízaga, MD, PhD∗,
- Yolanda Benito, DCS, DVM∗,
- Ana González-Mansilla, MD, PhD∗,
- Alicia Barrio, DCS, MBiol∗,
- Daniel Rodríguez-Pérez, PhD†,
- Pablo Martínez-Legazpi, MEng, PhD‡ and
- Francisco Fernández-Avilés, MD, PhD∗
- ∗Department of Cardiology, Hospital General Universitario Gregorio Marañón, Instituto de Investigación Sanitaria Gregorio Marañón, and Facultad de Medicina, Universidad Complutense de Madrid, Madrid, Spain
- †Department of Mathematical Physics and Fluids, Facultad de Ciencias, Universidad Nacional de Educación a Distancia, Madrid, Spain
- ‡Mechanical and Aerospace Engineering Department, University of California San Diego, San Diego, California
- ↵∗Reprint requests and correspondence:
Dr. Raquel Yotti, Department of Cardiology, Hospital General Universitario Gregorio Marañón, Dr. Esquerdo 46, 28007 Madrid, Spain.
Background Systemic arterial load impacts the symptomatic status and outcome of patients with calcific degenerative aortic stenosis (AS). However, assessing vascular properties is challenging because the arterial tree’s behavior could be influenced by the valvular obstruction.
Objectives This study sought to characterize the interaction between valvular and vascular functions in patients with AS by using transcatheter aortic valve replacement (TAVR) as a clinical model of isolated intervention.
Methods Aortic pressure and flow were measured simultaneously using high-fidelity sensors in 23 patients (mean 79 ± 7 years of age) before and after TAVR. Blood pressure and clinical response were registered at 6-month follow-up.
Results Systolic and pulse arterial pressures, as well as indices of vascular function (vascular resistance, aortic input impedance, compliance, and arterial elastance), were significantly modified by TAVR, exhibiting stiffer vascular behavior post-intervention (all, p < 0.05). Peak left ventricular pressure decreased after TAVR (186 ± 36 mm Hg vs. 162 ± 23 mm Hg, respectively; p = 0.003) but remained at >140 mm Hg in 70% of patients. Wave intensity analysis showed abnormally low forward and backward compression waves at baseline, increasing significantly after TAVR. Stroke volume decreased (−21 ± 19%; p < 0.001) and correlated with continuous and pulsatile indices of arterial load. In the 48 h following TAVR, a hypertensive response was observed in 12 patients (52%), and after 6-month follow-up, 5 patients required further intensification of discharge antihypertensive therapy.
Conclusions Vascular function in calcific degenerative AS is conditioned by the upstream valvular obstruction that dampens forward and backward compression waves in the arterial tree. An increase in vascular load after TAVR limits the procedure’s acute afterload relief.
Calcific degenerative aortic valve stenosis (AS) has become endemic in Western countries. For a given degree of valve obstruction, systemic arterial properties may impact the symptomatic status and outcome of these patients (1–3). In AS, left ventricular (LV) afterload is abnormally high because concentric remodeling and hypertrophy are insufficient to compensate for the additive effects of valvular obstruction and vascular load (4). Thus, vascular stiffness may be a source of LV systolic and diastolic dysfunctions in patients with moderate degrees of valve obstruction (3). This mechanism helps explain abnormally high morbidity and mortality rates in patients with AS for whom classical obstruction indices fail to predict outcomes (2).
Characterizing intrinsic properties of the arterial tree remains particularly challenging in AS because of the difficulties of uncoupling valvular and vascular functions in vivo (5). Acute and chronic interventions on either compartment cause reciprocal changes in the other. For instance, changes in vascular resistance caused by vasodilators (6,7) and exercise (8) induce significant modifications in valve hemodynamics. Likewise, valve interventions may acutely impact arterial function (9).
Although attempts have been made to quantify vascular load in AS noninvasively (2,4), a rigorous quantification of arterial hemodynamics entails simultaneous measurements of central aortic pressure and flow (10). Use of this invasive approach in a small number of subjects has suggested that steady and pulsatile loads are increased in symptomatic degenerative calcific AS, particularly during exercise (8). However, measurements of vascular load might be conditioned by upstream valvular obstruction.
This study was designed to characterize the interaction between valvular and vascular function in patients with calcific degenerative AS. We hypothesized that transcatheter aortic valve replacement (TAVR) offers a useful clinical model of isolated valvular intervention to unmask underlying valvular-vascular interactions of AS. Therefore, we analyzed the acute changes induced by TAVR to understand how valve obstruction impacts vascular function, using state-of-the-art methods, including frequency domain and wave intensity analyses (WIA) of high-fidelity data.
We studied 23 consecutive patients with severe symptomatic calcific degenerative AS undergoing TAVR (Table 1). Patients were either in sinus rhythm or permanent right ventricular (RV) pacing (n = 3). No patient had significant aortic regurgitation (AR), and 7 patients had an ejection fraction of ≤45%. Low-gradient AS (mean: <40 mm Hg) was present in 9 patients and concomitant low-flow (stroke volume [SV] index of <35 ml/m2) in 3 patients. Sixteen patients (74%) had a pre-procedural diagnosis of hypertension requiring pharmacotherapy. Antihypertensive agents were withheld 12 h before the procedure. After TAVR, patients were initially kept on their pre-procedural antihypertensive therapy. The local Institutional Review Board approved the study protocol and all subjects provided written informed consent.
Study protocol and follow-up
Procedures were performed using the femoral approach under local anesthesia and conscious sedation with low doses of midazolam (2 to 5 mg, intravenous) and fentanyl (2 μg/kg, intravenous); additional boluses (1 mg and 50 μg, respectively) were used if necessary to maintain patient comfort during the procedure. Special care was taken to ensure a constant level of sedation during pre- and post-procedural measurements. A pacing wire and a thermodilution Swan-Ganz catheter were placed in the RV and in the main pulmonary artery, respectively. The self-expanding valve (Corevalve, Medtronic, Inc., Minneapolis, Minnesota) transfemoral implantation procedure (11) was successful in all patients. Mild residual AR was present in 10 patients (grade 1 in 9 patients and grade 2 in 1 patient). Aortic and LV pressures were simultaneously recorded before and after TAVR, using fluid-filled catheters. Aortic valve areas were calculated using the Gorlin formula.
High-fidelity pressure and flow velocity were recorded simultaneously at the ascending aorta by using a 0.014-inch-diameter wire (Combowire, Volcano Corp., San Diego, California) under stable hemodynamic conditions (<10% variation in mean blood pressure [BP] during ≥10 min before and ≥30 min after TAVR). To minimize artifacts within the region of pressure recovery, the wire was introduced though a 6-F multipurpose guiding catheter placed in the ascending aorta ∼5 cm above the aortic annulus (Central Illustration). The Doppler and micromanometry sensors located at the wire’s tip were advanced approximately 1 cm out of the guiding catheter before data recording. After TAVR, the pressure-velocity wire was reinserted, matching the tip’s position fluoroscopically stored in the baseline study. The pressure signal was balanced against the fluid-filled guiding catheter. Signals were recorded for at least 1 minute during sinus rhythm and then during RV pacing at 20 beats/min above intrinsic baseline heart rate in all patients before and after TAVR. In patients with permanent RV pacing or those who were developing new-onset complete atrioventricular or left branch bundle block (n = 9), we used pacing signals before and after TAVR. High-fidelity pressure, flow velocity, and electrocardiogram signals were digitally stored at 200 Hz.
Comprehensive Doppler electrocardiogram examinations were performed immediately before and <24 h after TAVR, using broadband 2.0- to 4.0-MHz matrix and volumetric transducers on a Vivid-7 or a Vivid-9 system (General Electric Healthcare, Little Chalfont, United Kingdom). Cuff BP was monitored hourly during the first 48 h and then every 8 h until discharge. Hypertensive response after TAVR was defined (12) in the presence of 1 of the following: 1) sustained (>48-h) systolic pressure >140 mm Hg or diastolic pressure >90 mm Hg not present before; 2) need for a >2-fold increase in the dosage of an antihypertensive drug to achieve BP control; or 3) incorporation of an additional antihypertensive drug to the pre-procedural regimen. Patients underwent clinical follow-up, blinded to the results of vascular hemodynamics, every 3 months during the 6 months’ post-procedure.
Invasive data processing and analysis
Volumetric flow rate (ml/s) was calculated from linear flow velocity measurements (cm/s) by means of a calibration constant (cm2) obtained as K = SV/TVI, where TVI represents the time-velocity integral and SV is the simultaneously obtained thermodilution SV. Beats were selected for analysis if peak ascending aortic pressure exhibited variation of <10 mm Hg over the interval examined and the flow velocity waveform was stable and periodic (8). For each hemodynamic run, 13 beats (range: 5 to 19 beats) underwent digital low-pass (50-Hz) filtering and ensemble averaging (Figure 1) (8,13). The aortic input impedance spectrum was derived using Fourier decomposition of the pressure and velocity signals up to 10 Hz (10). Respective pressure and flow moduli at each harmonic were used to derive the impedance (Z) moduli. Characteristic impedance (Zc) was calculated as the average of Z moduli above 4 Hz, excluding outlier values of >3 times the median. Because this method is highly sensitive to signal noise, we additionally calculated Zc from wave speed, the latter measured in the time domain from the early P–Q linear relationship, as used for measuring wave velocity (Online Appendix). Correlation and agreement for both methods for measuring Zc were r = 0.67 and ric = 0.59, respectively (pooled before and after TAVR data). The augmentation index was computed as the difference between the maximum and minimum values of Z components >3 Hz. We calculated the distance to the reflecting site by the quarter-wavelength relationship (14), as well as by WIA (r = 0.51 and ric = 0.40 between methods [Online Appendix]). Arterial compliance (C) was calculated using the pulse pressure method (15), exponential decay, and diastolic area methods (10) (r > 0.92 and ric ≥ 0.90, among all methods). We calculated effective arterial elastance as: 1) the ratio between end systolic pressure (obtained from the fluid-filled LV pressure catheter) and SV (Ea); and 2) the ratio between systemic vascular resistance and the cardiac period (EaR; r = 0.95 and ric = 0.67 between methods) (16,17).
WIA is a well-established method used to assess arterial hemodynamics (18); its foundations define pressure and velocity waveforms as the summation of successive infinitesimal waves that propagate through vessels (18). Arterial waves can originate either from the LV (forward traveling) or from peripheral vasculature reflections (backward traveling). Waves are further classified by their effect on pressure as compression (increased pressure) or expansion w (decreased pressure) waves. We used the ensemble-averaged pressure and velocity signals to derive the rates of change of aortic pressure (dP/dt) and velocity (dU/dt) (Figure 1, Online Appendix). It has been proposed that changes in aortic pressure can be attributed not only to forward or backward wave motion but also to changes in aortic volume (19). Because we anticipated a potential effect of TAVR on aortic pressure and volume, we also performed WIA taking reservoir pressure effect into account (Online Figures 1 and 2) (19). All invasive data were analyzed using custom-built algorithms (Matlab; Mathworks, Natick, Massachusetts), and results for 3 to 5 hemodynamic runs were averaged for each patient.
Noninvasive valvulo-arterial impedance (ZVA) was calculated as: , where SBP is the cuff systolic BP, MG is the Doppler-derived mean transvalvular pressure gradient, and SVInoninv is the noninvasive SV index (SVI) measured by cross-sectional echocardiography and pulsed-wave Doppler (2).
Differences between pre- and post-TAVR hemodynamic data were analyzed by paired t tests. Responses between groups were compared using unpaired t tests. Correlation between quantitative variables was analyzed using the linear Pearson correlation coefficient (r), and 95% confidence interval (CI) for the fitting was plotted. The intraclass correlation coefficient (ric, absolute agreement) was used to compare different methods. Outcome analysis was performed by binary logistic regression models, accounting for improvement in New York Heart Association (NYHA) functional class at follow-up. SVI pre- and post-TAVR and its changes were entered separately in these models, adjusting for age and pre-implantation functional class. Because of the risk of overfitting in small samples, overall performance of the model was calculated using 1,000 bootstrap resamples to estimate the C index (20,21). Values of p < 0.05 were considered significant.
Indices of aortic stenosis and systemic hemodynamics
The large reduction in the transvalvular pressure gradient caused by TAVR was followed by significant increases in systolic, mean, and pulse systemic arterial pressure values (Table 2). Consequently, LV peak systolic pressure decreased by only a mean of 10% (186 ± 36 mm Hg vs. 162 ± 23 mm Hg, respectively; p = 0.003) and remained >140 mm Hg in 70% of patients, varying widely among patients (Figure 2). After TAVR, SVI (41 ± 8 ml/m2 vs. 33 ± 10 ml/m2, respectively; p < 0.001) and cardiac index (3.3 ± 0.8 l/min/m2 vs. 2.8 ± 1.1 l/min/m2, respectively; p <0.001) decreased (Figure 2). Patients with and without residual aortic regurgitation showed no significant differences in post-procedural LV end-diastolic pressure (31 ± 9 mm Hg vs. 26 ± 10 mm Hg, respectively; p = 0.22).
Systemic vascular load
A significant increase in systemic vascular resistance, Ea, and the first 3 harmonic frequencies of Z were observed after TAVR (Table 3, Central Illustration). The augmentation index and wave speed velocity increased as well, whereas C decreased (Table 3). The amount of decrease in C after TAVR was inversely related to baseline systolic BP (r = −0.72; p < 0.0001). SVI post-TAVR was strongly related to indices of continuous and pulsatile arterial load (Figure 3). Changes in SVI and arterial load indices (C, Ea, systemic vascular resistance, and Zc) were not significantly different among patients who did and did not require RV pacing after the procedure (p ≥ 0.1 for all).
TAVR was followed by a significant increase in forward compression waves (FCW) and backward compression waves (BCW) (Table 3, Central Illustration), whereas expansion waves did not change significantly. Pulse pressure and Zc increased, as measured by both the conventional and reservoir approach methods (Online Table 1); the reflection coefficient increased following TAVR, whereas the distance to reflection was only found to decrease by using the reservoir method. Pulse pressure correlated directly with compression waves (r = 0.53 and r = 0.62 for peak FCWs and BCWs, respectively), directly with the backward expansion wave (r = 0.70), and inversely with the forward expansion wave (r = −0.65; p < 0.0001 for all, pooled data and reservoir approach). The Zva did not change significantly with TAVR (4.1 ± 1.2 mm Hg/ml/m2 vs. 3.9 ± 1.4 mm Hg/ml/m2, respectively; p = 0.59).
In the 48 h following TAVR, a hypertensive response was observed in 12 patients (52%); 10 patients required intensification of their antihypertensive therapy and 1 initiation of treatment. During 6 months of follow-up, 5 patients had their discharge antihypertensive therapy intensified, whereas no patient had reduced doses of these drugs. NYHA functional class did not improve in 14 patients (61%). Improvement in functional class after TAVR was directly related to post-procedural SVI (odds-ratio [OR]: 2.8 [95% CI: 1.1 to 7.3] per 5 ml; bootstrapped C index: 0.67; p = 0.03) and inversely to the fall in SVI observed after TAVR (OR: 0.3 [95% CI: 0.1 to 0.9] per 5 ml; p = 0.05), whereas it was not related to pre-TAVR SVI (p = 0.4).
The present study clarifies important aspects of vascular adaptation to calcific degenerative AS. Using WIA, we demonstrated that valvular obstruction blunts the conversion of LV ejection blood momentum into FCWs in the arterial system. Dampened FCWs are reflected as abnormally low BCWs at the aortic bifurcation sites, and both effects result in low systolic and pulse arterial pressures. This situation changes acutely after TAVR, demonstrating that the characterization of systemic vascular properties in AS is conditioned by the upstream obstruction. The relief of the outflow obstruction immediately raises FCWs and BCWs, increasing arterial pressures and vascular impedance and induces a stiffer vascular behavior. In our study, the augmented vascular load correlated with post-procedural SVI. Although this study was limited to a small sample size, we found an inverse relationship between the procedure’s mid-term clinical benefit and the change in SVI observed post-TAVR.
Vascular tree in degenerative calcific AS
Noninvasive (4) and mathematical (5) methods have described abnormally high steady and pulsatile components of systemic arterial load in patients with degenerative calcific AS. However, few studies have analyzed the status of intrinsic vascular properties in AS invasively. Laskey et al. (8) compared 18 patients with symptomatic degenerative calcific AS to 11 younger control subjects and found higher vascular resistance and impedance and reduced arterial compliance in patients with AS. Differences between groups became particularly evident during exercise (8). However, our study’s results suggest that these observations should be interpreted cautiously. By analyzing the acute response to TAVR, we showed that valve stenosis per se influences all metrics characterizing the arterial tree. Noticeably, classical values of vascular function obtained in our study pre-TAVR did not differ from previously reported values in age-matched hypertensive populations (22). However, WIA showed that compression and expansion waves in AS are much lower than previously reported normal values (23). We found that immediately after TAVR, the transmission of blood momentum to the arterial system improves, increasing FCWs. Stronger FCWs are reflected as stronger BCWs traveling toward the LV. Both effects raise mean, systolic, and pulse arterial pressure levels.
Our results show that after TAVR, the vascular tree exhibits a stiffer behavior. This paradoxical effect of rising continuous and pulsatile vascular load after LV outflow relief was described previously (9). In patients undergoing percutaneous aortic valvuloplasty, valvular-vascular interaction follows the properties of complementarity (both compartments contribute additively to afterload) and competitiveness (one compartment cannot be lowered without raising the other one) (9,24). More recently, this interaction was confirmed in AS patients undergoing vasodilator pharmacological interventions (6). Our study demonstrates the negative impact this vascular response exerts on global hemodynamics of patients undergoing TAVR. Although we did not repeat invasive studies during follow-up, the relatively large proportion of patients requiring antihypertensive therapy scaling during follow-up suggests that our acute observations are not acute phase transients. Similar observations of persistent hypertension have been reported after TAVR (12) and surgical valve replacement (25).
Changes in the tone of large conduction arteries and arterioles are probably responsible for the observed changes in pulsatile vascular load after TAVR. We know vasoconstriction in arteriolar vessels reduces arterial compliance (14). The viscoelastic properties of large conductance arteries also may be responsible for this observation. Due to the nonlinearity of viscoelastic strain of the large conductance arteries, acute changes in the pressure-mediated deformation post-TAVR may also induce stiffer behavior of the vascular tree (26).
Vascular tree and outcome in AS
Indirect evidence has emphasized the complementary impact of arterial hemodynamics on the symptomatic status and outcome of patients with AS, both before (2,4) and after (27) valve intervention. The “double loaded” hypothesis integrates these additive effects of valvular and vascular loads. On this basis, the ZVA index has been found to correlate with SV and outcome (3,28). However, in our study, ZVA did not capture the hemodynamic changes observed with TAVR. The fact that ZVA did not improve acutely probably relates to its sensitivity to both the valvular and the vascular compartments, which are competitively modified by therapy.
A well-known risk factor of cardiovascular morbidity and mortality, especially in elderly patients (29), hypertension has been associated with worse outcomes in patients who undergo TAVR (30). However, a hypertensive response after TAVR has also been associated with a better prognosis (12). In a previous study, higher BP after TAVR was related to higher SV and was attributed to an acute improvement of LV function; patients with stable BP after TAVR experienced, on average, a reduction in post-procedural SV (12). Similarly, our study suggests that a post-procedural reduction in SV is related to absence of clinical improvement. However, we have shown that the acute hypertensive response after TAVR is caused by increased vascular load rather than improved LV systolic function, so it should be promptly identified and treated.
Because no striking changes in chamber systolic volume are expected during TAVR, the observation of post-procedural increased arterial load suggests that the hemodynamic benefits of valvular replacement on LV systolic wall stress may be lower than expected, particularly in patients with relatively low transvalvular pressure gradients. Although peak LV pressure decreases after TAVR, it frequently remained higher than normal. Increased post-procedural vascular load may explain why patients with paradoxically low-flow low-gradient AS fail to improve values of N-terminal prohormone B-type natriuretic peptide by 1 year after TAVR (31) and have a higher mortality than patients with normal flow (32). Further large-scale studies are necessary to address the predictors of LV systolic stress improvement. Nevertheless, in view of our data and those of others (6), intense medical therapy is recommended in hypertensive patients with calcific degenerative AS, regardless of whether they finally do or do not undergo valve replacement.
The flow acquisition system measures aortic flow velocity by using a very small Doppler sample volume. Therefore, signals may sometimes be noisy in highly turbulent flows, as in AS, and not account for the average flow velocity for the full cross-section of the aorta where measurements are obtained. The geometry of the Corevalve prosthesis can modify the local mechanical properties of the arterial wall in the aortic root. For this reason, we acquired the invasive pressure and flow rate/velocity signals 5 cm distal to the aortic annulus, searching for the highest velocities at this point, attempting to minimize the prosthesis’ local effects. To avoid these issues, we selected data with the highest quality available and implemented filtering and ensemble averaging to increase the signal-to-noise ratio. However, we cannot exclude the fact that residual high-frequency noise may account for the relatively high Zc values that were measured. Although a stable conscious sedation level was achieved in all cases, a certain vascular tone modification can be expected for sedating drugs. Similarly, some degree of vascular changes could be caused by adaptation to acute procedure-related myocardial injury.
We studied an elderly and high-risk AS group; therefore, the vascular hemodynamics and response to TAVR could be different in younger patients. Functional improvement was only assessed using NYHA functional classification; other tools such as the 6-min walk test or quality-of-life questionnaires would have increased the sensitivity to detect functional improvement. The small sample size was designed to analyze the mechanistic changes in vascular load. Thus, subgroup analyses need to be interpreted cautiously, and hard clinical endpoints could not be analyzed. With the small sample size, we also could not address the impact of potential confounders such as degree of mitral regurgitation. Large-scale clinical studies are necessary to rule out a potential acute rebound effect post-intervention and confirm that post-TAVR measurements accurately account for the true arterial load once the stenotic damping effect has been alleviated.
Because valvular and vascular loads are tightly coupled in AS, upstream obstruction can influence the measurements of arterial properties. Low arterial FCWs and BCWs caused by valvular stenosis produce the hallmark signs of arterial hemodynamics in AS. Relief of valvular obstruction with TAVR acutely increases compression waves, causing the arterial tree to operate at a higher pressure and therefore increasing the vascular load. This phenomenon impacts the post-procedural acute hemodynamic benefits of TAVR.
COMPETENCY IN MEDICAL KNOWLEDGE: Relief of AS raises forward and backward compression waves, increasing arterial pressure and both the steady and pulsatile components of systemic arterial load.
COMPETENCY IN PROCEDURAL SKILLS: The increased post-procedural systemic vascular load should be promptly treated when patients with AS undergo TAVR, particularly when the transvalvular pressure gradient is low.
TRANSLATIONAL OUTLOOK: Larger prospective studies are needed to define the prognostic implications of changes in systemic vascular load that immediately follow TAVR.
The author thank all members of the staff of the Echocardiography and Catheterization Laboratories of the Hospital General Universitario Gregorio Marañón for their support for patient recruitment and data collection.
For expanded Methods and Results sections, including a supplemental table and figures, please see the online version of this article.
This study was supported by Instituto de Salud Carlos III, Ministerio de Economía y Competitividad, Spain, grants PIS09/02602, PIS012/02878, RD12/0042, CM12/00273 (to Dr. Perez del Villar), and CM11/00221 (to Dr. Mombiela). Drs. Mombiela, González-Mansilla, and del Villar were partially supported by grants from the Fundación para Investigación Biomédica Gregorio Marañón, Spain. Dr. Martínez-Legazpi was supported by U.S. National Institutes of Health grant 1R21 HL108268-01. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
This work was presented in part at the Scientific Sessions of the American Heart Association, 2012, Los Angeles, California, November 4 to 7; abstract A15474.
- Abbreviations and Acronyms
- aortic stenosis
- backward compression wave
- systemic arterial elastance
- forward compression wave
- stroke volume index
- transcatheter aortic valve replacement
- wave intensity analysis
- characteristic impedance
- Received August 5, 2014.
- Revision received October 13, 2014.
- Accepted October 21, 2014.
- American College of Cardiology Foundation
- Bermejo J.,
- Odreman R.,
- Feijoo J.,
- Moreno M.M.,
- Gomez-Moreno P.,
- Garcia-Fernandez M.A.
- Hachicha Z.,
- Dumesnil J.G.,
- Bogaty P.,
- Pibarot P.
- Hachicha Z.,
- Dumesnil J.G.,
- Pibarot P.
- Briand M.,
- Dumesnil J.G.,
- Kadem L.,
- et al.
- Garcia D.,
- Barenbrug P.J.,
- Pibarot P.,
- et al.
- Eleid M.F.,
- Nishimura R.A.,
- Sorajja P.,
- Borlaug B.A.
- Jimenez-Candil J.,
- Bermejo J.,
- Yotti R.,
- et al.
- Laskey W.K.,
- Kussmaul W.G. III.,
- Noordergraaf A.
- Shim Y.,
- Hampton T.G.,
- Straley C.A.,
- et al.
- Laskey W.K.,
- Parker H.G.,
- Ferrari V.A.,
- Kussmaul W.G.,
- Noordergraaf A.
- Grube E.,
- Schuler G.,
- Buellesfeld L.,
- et al.
- Perlman G.Y.,
- Loncar S.,
- Pollak A.,
- et al.
- Davies J.E.,
- Sen S.,
- Broyd C.,
- et al.
- Nichols W.W.,
- O'Rourke M.,
- Vlachopoulos C.
- Chemla D.,
- Hebert J.L.,
- Coirault C.,
- et al.
- Segers P.,
- Stergiopulos N.,
- Westerhof N.
- Sunagawa K.,
- Maughan W.L.,
- Sagawa K.
- R Core Team
- Harrell F.E.
- Mitchell G.F.,
- Lacourciere Y.,
- Ouellet J.P.,
- et al.
- Milnor W.R.,
- Bertram C.D.
- Aronow W.S.,
- Fleg J.L.,
- Pepine C.J.,
- et al.
- Thomas M.,
- Schymik G.,
- Walther T.,
- et al.
- Le Ven F.,
- Freeman M.,
- Webb J.,
- et al.